The image processing handbook 6ed.

Edition Russ
J.C. pdf download

https://ebookname.com/product/the-image-processing-handbook-6ed-
edition-russ-j-c/

Get the full ebook with Bonus Features for a Better Reading Experience on ebookname.com
Instant digital products (PDF, ePub, MOBI) available
Download now and explore formats that suit you...

The Image Processing Handbook 6th Edition Thomas

Klinger

https://ebookname.com/product/the-image-processing-handbook-6th-
edition-thomas-klinger/

Image Processing The Fundamentals 2nd 2nd Edition Maria

Petrou

https://ebookname.com/product/image-processing-the-
fundamentals-2nd-2nd-edition-maria-petrou/

Multivariate Image Processing 1st Edition Jocelyn

Chanussot

https://ebookname.com/product/multivariate-image-processing-1st-
edition-jocelyn-chanussot/

Avoiding and Resolving Disputes A Short Guide for

Architects 1st Edition Bart Kavanagh (Author)

https://ebookname.com/product/avoiding-and-resolving-disputes-a-
short-guide-for-architects-1st-edition-bart-kavanagh-author/
Electronic Instrumentation 2006th Edition Regtien

https://ebookname.com/product/electronic-instrumentation-2006th-
edition-regtien/

A Manual of Egyptian Pottery Naqada III Middle Kingdom

1, revised Edition Anna Wodzinska

https://ebookname.com/product/a-manual-of-egyptian-pottery-
naqada-iii-middle-kingdom-1-revised-edition-anna-wodzinska/

Creating E Learning Games with Unity 2015th Edition

David Horachek

https://ebookname.com/product/creating-e-learning-games-with-
unity-2015th-edition-david-horachek/

Grokking Algorithms Second Edition Aditya Y Bhargava

https://ebookname.com/product/grokking-algorithms-second-edition-
aditya-y-bhargava/

Science of the Sages Scientists Encountering Nonduality

from Quantum Physics to Cosmology to Consciousness
Robert Wolfe

https://ebookname.com/product/science-of-the-sages-scientists-
encountering-nonduality-from-quantum-physics-to-cosmology-to-
consciousness-robert-wolfe/
How to change the world reflections on Marx and Marxism
Hobsbawm

https://ebookname.com/product/how-to-change-the-world-
reflections-on-marx-and-marxism-hobsbawm/
 The
IMAGE
PROCESSING
Handbook
Sixth Edition

John C. Russ
 The

IMAGE
PROCESSING
Handbook
Sixth Edition
 The

IMAGE
PROCESSING
Handbook
Sixth Edition
John C. Russ
North Carolina State University
Materials Science and Engineering Department
Raleigh, North Carolina

Boca Raton London New York

CRC Press is an imprint of the

Taylor & Francis Group, an informa business
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
© 2011 by Taylor and Francis Group, LLC
CRC Press is an imprint of Taylor & Francis Group, an Informa business

No claim to original U.S. Government works

Printed in the United States of America on acid-free paper

10 9 8 7 6 5 4 3 2 1

International Standard Book Number-13: 978-1-4398-4063-4 (Ebook-PDF)

This book contains information obtained from authentic and highly regarded sources. Reasonable efforts have been
made to publish reliable data and information, but the author and publisher cannot assume responsibility for the valid-
ity of all materials or the consequences of their use. The authors and publishers have attempted to trace the copyright
holders of all material reproduced in this publication and apologize to copyright holders if permission to publish in this
form has not been obtained. If any copyright material has not been acknowledged please write and let us know so we may
rectify in any future reprint.

Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmitted, or uti-
lized in any form by any electronic, mechanical, or other means, now known or hereafter invented, including photocopy-
ing, microfilming, and recording, or in any information storage or retrieval system, without written permission from the
publishers.

For permission to photocopy or use material electronically from this work, please access www.copyright.com (http://
www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923,
978-750-8400. CCC is a not-for-profit organization that provides licenses and registration for a variety of users. For
organizations that have been granted a photocopy license by the CCC, a separate system of payment has been arranged.

Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for
identification and explanation without intent to infringe.
Visit the Taylor & Francis Web site at
http://www.taylorandfrancis.com
and the CRC Press Web site at
http://www.crcpress.com
Contents

Acknowledgments. . . . . . . . . . . . . . . . . . . xi
Introduction . . . . . . . . . . . . . . . . . . . . . . xiii
Author . . . . . . . . . . . . . . . . . . . . . . . . . xvii

1 Acquiring Images . . . . . . . . . . . . . . . . 1
Human reliance on images for information............ 1
Video cameras................................................... 6
CCD cameras.................................................... 7
Camera artifacts and limitations..........................13
Color cameras..................................................15
Camera resolution.............................................18
Focusing.......................................................... 20
Electronics and bandwidth limitations...................21
Pixels...............................................................24
Gray scale resolution........................................ 26
Noise............................................................. 28
High depth images........................................... 30
Color imaging...................................................34
Digital camera limitations.................................. 42
Color spaces................................................... 42
Color correction................................................52
Color displays...................................................54
Image types..................................................... 56
Range imaging................................................ 58
Multiple images............................................... 64
Stereoscopy..................................................... 69
Imaging requirements........................................ 77

v
© 2011 by Taylor & Francis Group, LLC
 2 Human Vision. . . . . . . . . . . . . . . . . . 85
What we see and why...................................... 85
Recognition..................................................... 88
Technical specs................................................ 92
Acuity..............................................................97
What the eye tells the brain..............................101
Spatial comparisons........................................103
Local to global hierarchies................................107
It’s about time................................................. 113
The third dimension.......................................... 118
How versus what............................................. 121
Seeing what isn’t there, and vice versa...............122
Image compression..........................................125
A world of light...............................................126
Size matters....................................................129
Shape (whatever that means)............................132
Context..........................................................133
Arrangements must be made.............................135
Seeing is believing.......................................... 137
So in conclusion..............................................139

3 Printing and Storage. . . . . . . . . . . . 141

Printing.......................................................... 141
Dots on paper.................................................145
Color printing.................................................150
Printing hardware............................................156
Film recorders................................................. 161
Other presentation tools...................................162
File storage.....................................................163
Storage media................................................164
Magnetic recording.........................................166
Databases for images......................................167
Browsing and thumbnails.................................. 174
Lossless coding................................................ 178
Reduced color palettes.....................................183
JPEG compression...........................................184
Wavelet compression.......................................187
Fractal compression.........................................192
Digital movies.................................................194

4 Correcting Imaging Defects . . . . . . . 199

Contrast expansion......................................... 200
Noisy images................................................ 205
Neighborhood averaging................................ 208
Neighborhood ranking.................................... 214
Other neighborhood noise reduction methods.....226

vi Contents
© 2011 by Taylor & Francis Group, LLC
 Defect removal, maximum entropy, and
maximum likelihood.........................................232
Nonuniform illumination...................................235
Fitting a background function............................238
Rank leveling..................................................244
Color images..................................................248
Non-planar views............................................250
Computer graphics..........................................252
Geometric distortion........................................254
Alignment.......................................................256
Interpolation...................................................261
Morphing.......................................................265

5 Image Enhancement in the Spatial

Domain. . . . . . . . . . . . . . . . . . . . . . 269
Contrast manipulation......................................270
Histogram equalization.................................... 274
Local equalization...........................................279
Laplacian.......................................................283
Derivatives......................................................293
Finding edges with gradients............................296
More edge detectors...................................... 306
Texture........................................................... 312
Fractal analysis............................................... 317
Implementation notes....................................... 319
Image math.................................................... 319
Subtracting images..........................................320
Multiplication and division................................323
Principal components analysis...........................325
Other image combinations...............................331

6 Processing Images in Frequency

Space. . . . . . . . . . . . . . . . . . . . . . . 337
About frequency space....................................337
The Fourier transform.......................................338
Fourier transforms of simple functions.................341
Frequencies and orientations.............................345
Preferred orientation........................................350
Texture and fractals.........................................351
Isolating periodic noise....................................356
Selective masks and filters................................361
Selection of periodic information.......................364
Convolution....................................................370
Deconvolution.................................................372
Noise and Wiener deconvolution......................378
Template matching and correlation....................385
Autocorrelation...............................................391

Contents vii
© 2011 by Taylor & Francis Group, LLC
 7 Segmentation and Thresholding. . . .395
Thresholding...................................................395
Automatic settings...........................................398
Multiband images.......................................... 403
Two-dimensional thresholds.............................. 405
Multiband thresholding................................... 408
Thresholding from texture.................................. 411
Multiple thresholding criteria............................. 414
Textural orientation.......................................... 415
Region boundaries.......................................... 419
Conditional histograms.....................................426
Boundary lines................................................427
Contours....................................................... 430
Image representation.......................................432
Other segmentation methods............................436
The general classification problem.................... 440

8 Processing Binary Images. . . . . . . . 443

Boolean operations........................................ 443
Combining Boolean operations........................ 446
Masks........................................................... 450
From pixels to features.....................................452
Boolean logic with features...............................457
Selecting features by location...........................461
Double thresholding........................................ 466
Erosion and dilation........................................ 468
Opening and closing.......................................471
Isotropy..........................................................473
Measurements using erosion and dilation...........478
Extension to gray scale images.........................481
Morphology neighborhood parameters............. 482
Examples of use..............................................484
Euclidean distance map.................................. 488
Watershed segmentation..................................491
Ultimate eroded points.....................................494
Skeletons........................................................498
Boundary lines and thickening......................... 503
Combining skeleton and EDM.......................... 506

9 Global Image Measurements. . . . . . 511

Global measurements and stereology................ 511
Surface area................................................... 516
ASTM Grain Size............................................. 521
Multiple types of surfaces.................................523
Length............................................................525
Thickness........................................................527
Sampling strategies.........................................530

viii Contents
© 2011 by Taylor & Francis Group, LLC
 Determining number........................................532
Curvature, connectivity, and the Disector............535
Anisotropy and gradients.................................538
Size distributions.............................................542
Classical stereology (unfolding).........................543

10 Feature-Specific Measurements . . . . 547

Brightness measurements..................................547
Determining location........................................556
Orientation.....................................................559
Neighbor relationships.....................................562
Alignment.......................................................567
Counting........................................................ 574
Special counting procedures.............................579
Feature size....................................................584
Circles and ellipses..........................................587
Caliper dimensions..........................................589
Perimeter........................................................592

11 Characterizing Shape . . . . . . . . . . . 597

Describing shape............................................597
Dimensionless ratios.........................................599
Fractal dimension........................................... 604
Harmonic analysis........................................... 610
Topology........................................................620
Three dimensions.............................................623

12 Feature Recognition and

Classification. . . . . . . . . . . . . . . . . . 627
Template matching and cross-correlation............628
Parametric description......................................631
Decision points...............................................635
Multidimensional classification..........................639
Learning systems..............................................646
kNN and cluster analysis.................................652
Expert systems.................................................655
Neural nets.....................................................657
Syntactical models...........................................659

13 Tomographic Imaging. . . . . . . . . . . 661

More than two dimensions................................661
Volume imaging vs. sections.............................664
Basics of reconstruction....................................670
Algebraic reconstruction methods......................676
Maximum entropy...........................................679
Defects in reconstructed images........................681

Contents ix
© 2011 by Taylor & Francis Group, LLC
 Beam hardening............................................. 686
Imaging geometries.........................................691
Three-dimensional tomography..........................695
High-resolution tomography..............................701

14 3D Visualization . . . . . . . . . . . . . . . 707
Sources of 3D data..........................................707
Serial sections.................................................709
Optical sectioning........................................... 713
Sequential removal.......................................... 715
Stereo measurement........................................ 717
3D data sets...................................................722
Slicing the data set..........................................724
Arbitrary section planes...................................727
The use of color.............................................. 731
Volumetric display...........................................732
Stereo viewing................................................736
Special display hardware.................................739
Ray tracing..................................................... 741
Reflection.......................................................746
Surfaces.........................................................750
Multiply connected surfaces..............................754
Image processing in 3D...................................759
Measurements on 3D images............................763

15 Imaging Surfaces . . . . . . . . . . . . . . 767

Producing surfaces........................................... 767
Imaging surfaces by physical contact.................770
Noncontacting measurements...........................773
Microscopy of surfaces....................................777
Surface composition imaging............................782
Processing of range images..............................783
Processing of composition maps........................787
Data presentation and visualization...................788
Rendering and visualization.............................. 791
Analysis of surface data...................................796
Profile measurements....................................... 800
The Birmingham measurement suite................... 803
Topographic analysis and fractal dimensions..... 809

References . . . . . . . . . . . . . . . . . . . . . . 817

x Contents
© 2011 by Taylor & Francis Group, LLC
Acknowledgments

All of the image processing and the creation of the resulting figures included in this
book were performed on an Apple Macintosh® and/or a Sony VAIO® computer, using
Adobe Photoshop® CS4 with the Fovea Pro plug-ins. Many of the images were acquired
directly from various microscopes and other sources that provided digital output directly
to the computer. Others were captured using a variety of digital cameras (Sony, Nikon,
Canon, and others), and some were obtained using flat-bed and slide scanners (Nikon and
Epson), often from images supplied by colleagues and researchers. These are acknowl-
edged wherever the origin of an image could be determined. A few examples, taken
from the literature, are individually referenced.
The book was delivered to the publisher in digital form (on a writable DVD), without
intermediate hard copy, negatives or prints of the images, etc. Among other things,
this means that the author must bear full responsibility for typographical errors or
problems with the figures. Every effort has been made to show enlarged image frag-
ments that will reveal pixel-level detail when it is important. The process has also
forced me to learn more than I ever hoped to know about some aspects of publish-
ing technology! However, going directly from disk file to print also shortens the time
needed in production and helps to keep costs down, while preserving the full quality
of the images. Grateful acknowledgment is made of the efforts by the excellent edi-
tors at CRC Press to educate me and to accommodate the unusually large number of
illustrations in this book (more than 2000 figures and more than a quarter of a mil-
lion words).
Special thanks are due to Chris Russ (Reindeer Graphics Inc., Asheville, NC) who has
helped to program many of these algorithms and contributed invaluable comments, and
especially to Helen Adams, who has proofread many pages, endured many discussions
about ways to present information effectively, and provided the support (and the occa-
sional glass of wine) that make writing projects such as this possible.

John C. Russ
Raleigh, NC

xi
© 2011 by Taylor & Francis Group, LLC
Introduction

Image processing is used in a wide variety of applications, for two somewhat differ-
ent purposes:

1. improving the visual appearance of images to a human observer, including their

printing and transmission, and
2. preparing images for the measurement of the features and structures which they
reveal.

The techniques that are appropriate for each of these tasks are not always the same, but
there is considerable overlap. This book covers methods that are used for both tasks.
To do the best possible job, it is important to know about the uses to which the pro-
cessed images will be put. For visual enhancement, this means having some familiarity
with the human visual process and an appreciation of what cues the viewer responds
to in images. A chapter on human vision addresses those issues. It also is useful to
know about the printing or display process, since many images are processed in the
context of reproduction or transmission. Printing technology for images has advanced
significantly with the consumer impact of digital cameras, and up-to-date information
is provided.
The measurement of images is often a principal method for acquiring scientific data and
generally requires that features or structure be well defined, either by edges or unique
brightness, color, texture, or some combination of these factors. The types of measure-
ments that can be performed on entire scenes or on individual features are important in
determining the appropriate processing steps. Several chapters deal with measurement
in detail. Measurements of size, position, and brightness deal with topics that humans
largely understand, although human vision is not quantitative and is easily fooled. Shape
is a more difficult concept, and a separate chapter added in this edition summarizes a
variety of ways that shape may be described by numbers. Measurement data may be
used for classification or recognition of objects. There are several different strategies that
can be applied, and examples are shown.
It may help to recall that image processing, like food processing or word processing,
does not reduce the amount of data present but simply rearranges it. Some arrangements

xiii
© 2011 by Taylor & Francis Group, LLC
may be more appealing to the senses, and some may convey more meaning, but these
two criteria may not be identical nor call for identical methods.
This handbook presents an extensive collection of image processing tools, so that the
user of computer-based systems can both understand those methods provided in pack-
aged software and program those additions which may be needed for particular applica-
tions. Comparisons are presented for different algorithms that may be used for similar
purposes, using a selection of representative pictures from various microscopy tech-
niques, as well as macroscopic, remote sensing, and astronomical images. It is very
important to emphasize that the scale of the image matters very little to the techniques
used to process or analyze it. Microscopes that have a resolution of nm and telescopes
that produce images covering light years produce images that require many of the same
algorithms.
The emphasis throughout the book continues to be on explaining and illustrating meth-
ods so that they can be clearly understood, rather than providing dense mathematics.
With the advances in computer speed and power, tricks and approximations in search
of efficiency are less important, so that examples based on exact implementation of
methods with full precision can generally be implemented on desktop systems. The top-
ics covered are generally presented in the same order in which the methods would be
applied in a typical workflow.
For many years, in teaching this material to students I have described achieving mastery of
these techniques as being much like becoming a skilled journeyman carpenter. The num-
ber of distinct woodworking tools — saws, planes, drills, etc. — is relatively small, and
although there are some variations — slotted vs. Phillips-head screwdrivers, for example
— knowing how to use each type of tool is closely linked to understanding what it does.
With a set of these tools, the skilled carpenter can produce a house, a boat, or a piece of
furniture. So it is with image processing tools, which are conveniently grouped into only
a few classes, such as histogram modification, neighborhood operations, Fourier-space
processing, and so on, and can be used to accomplish a broad range of purposes. Visiting
your local hardware store and purchasing the appropriate tools do not provide the skills
to use them. Understanding their use requires practice, which develops the ability to visu-
alize beforehand what each will do. The same is true of the tools for image processing.
In revising the book for this new edition, I have again tried to respond to some of the
comments and requests of readers and reviewers. New chapters on the measurement
of images and the subsequent interpretation of the data were added in the second edi-
tion, and a section on surface images in the third. The fourth edition added the stereo-
logical interpretation of measurements on sections through three-dimensional structures
and the various logical approaches to feature classification. The fifth edition brought
expanded sections on deconvolution, extended dynamic range images, and multichannel
imaging, including principal components analysis. In this sixth edition, a new chapter on
the meaning of shape has been added, as well as additional material on imaging in more
than two dimensions. The sections on the ever-advancing hardware for image capture
and printing have been expanded and information added on the newest hardware and
software technologies.
As in past editions, I have resisted suggestions to put “more of the math” into the book.
There are excellent texts on image processing, compression, mathematical morphol-
ogy, etc., that provide as much rigor and as many derivations as may be needed. Many
of them are referenced here. But the thrust of this book remains teaching by example.

xiv Introduction
© 2011 by Taylor & Francis Group, LLC
Few people learn the principles of image processing from the equations. Just as we use
images to communicate ideas and to “do science,” so most of us use images to learn
about many things, including imaging itself. The hope is that by seeing and comparing
what various operations do to representative images, you will discover how and why to
use them. Then, if you need to look up the mathematical foundations, they will be easier
to understand.
A very real concern for everyone involved in imaging, particularly in scientific and foren-
sic fields, is the question of what constitutes proper and appropriate processing and what
constitutes unethical or even fraudulent manipulation. The short answer is that anything
that alters an image so as to create a false impression on the part of the viewer is wrong.
The problem with that answer is that it does not take into account the fact that different
viewers will tend to see different things in the image anyway, and that what constitutes
a false impression for one person may not for another.
The first rule is always to store a permanent copy of the original image along with rel-
evant data on its acquisition. The second rule is to carefully document whatever steps
are taken to process the image and generally to report those steps when the processed
image is published. Most scientific publications and the editors who review submitted
papers have become more aware in recent years of the ease with which image process-
ing can be performed and the dangers of inadequate documentation. For example, see
M. Rossner and K. M. Yamada (2004; J. Cell Biology) for that journal’s policy on image
ethics and examples of improper manipulation.
For forensic purposes, there is an additional responsibility to fully record the entire step-
by-step procedures that are used and to make sure that those methods are acceptable
in court according to the U.S. Supreme Court’s Daubert ruling (Daubert v. Merrell Dow
Pharmaceuticals (92-102), 509 U.S. 579, 1993), which generally means that not only are
the methods widely accepted by professionals, but also that they have been rigorously
tested and have known performance outcomes. In a forensic setting, there will often
be a need to explain a procedure, step by step, to a non-technical jury. This frequently
requires showing that the details obtained from the image are really present in the origi-
nal but only became visually evident with the processing.
Some procedures, such as rearranging features or combining them within a single image,
or differently adjusting the contrast of several images to make them appear more alike,
are clearly misleading and generally wrong. Some, such as using copy-and-paste to dupli-
cate a portion of an image, or selectively erasing portions of an image, are out-and-out
fraudulent. Even selective cropping of an image (or choosing which field of view to
record) can create a false impression.
The general guideline to be considered is that it is never acceptable to add anything to
an image, but it may be acceptable to suppress or remove some information if it makes
the remaining details more accessible, either visually for presentation and communication
or to facilitate measurement. Of course, the procedures used must be documented and
reported. Any of the procedures shown here may be appropriate in a particular instance.
But they can also be misused and should in any case never be used without understand-
ing and careful documentation. The heart of the scientific method is replicability. If
adequate information is provided on the processing steps applied and the original image
data are preserved, then the validity of the results can be independently verified.
An important but often overlooked concern is the need to avoid using programs that
alter the image without the user being aware of it. For example, carefully correcting the

Introduction xv
© 2011 by Taylor & Francis Group, LLC
colors in an image using Photoshop® and then placing it in PowerPoint® for presentation
will cause changes even on the same computer screen (as well as discarding pixels and
reducing resolution if copy-and-paste is used for the transfer). In addition, the image
may appear different on another computer monitor or when using a projector. Pasting an
image into Microsoft® Word will reduce the resolution and color or gray scale dynamic
range. This may not affect the printed document, which has less gamut than the com-
puter screen anyway, but the image cannot be subsequently retrieved from the document
in its original form. Saving an image with a lossy compression method such as jpeg will
discard potentially important information that cannot be recovered.
The reader is encouraged to use this book in concert with a real source of images and
a computer-based system and to freely experiment with different methods to deter-
mine which are most appropriate for his or her particular needs. Selection of image
processing tools to explore images when you don’t know the contents beforehand is
a much more difficult task than using tools to make it easier for another viewer or a
measurement program to see the same things you have discovered. It places greater
demand on computing speed and the interactive nature of the interface. But it particu-
larly requires that you become a very analytical observer of images. If you can learn
to see what the computer sees and predict what various algorithms will do, you will
become a better viewer and obtain the best possible images, suitable for further pro-
cessing and analysis.
To facilitate this hands-on learning, I have collaborated with my son, Chris Russ, to
write a companion book, Introduction to Image Processing and Analysis, which teaches
how to program these algorithms and create Adobe Photoshop compatible plug-ins that
implement the methods. The downloadable solutions to the book’s worked problems
can be used to apply the routines on either Macintosh or Windows computers. There are
additional links to downloadable plug-ins and trial program packages on my Web site at
http://www.DrJohnRuss.com.

xvi Introduction
© 2011 by Taylor & Francis Group, LLC
Author

In his fifty-year career as scientist

and educator, John Russ has used
image processing and analysis as
a principal tool for understanding
and characterizing the structure and
function of materials. Images from a
wide variety of devices — including
light and electron microscopes, x-ray
and neutron tomography, and more
— require computer processing and
measurement to extract the important
data. Much of Russ’ research work
has been concerned with the micro-
structure and surface topography of
metals and ceramics. He has received
funding for his research and teaching
from government agencies and from
industry. Although retired, Dr. Russ
is currently assisting in the establish-
ment of a new laboratory and program at North Carolina State University, which will be
the first in the nation to offer advanced degrees in Forensic Science and Engineering.
Familiarity with the algorithms and instruments led to Dr. Russ’ expertise being extended
to a much broader range of images — from astronomy to biomedical research to food
science to forensics. In addition to students in NCSU’s College of Engineering, Russ has
been on graduate student commitees and collaborated with faculty in textiles, pulp
and paper products, veterinary medicine, microbiology, food science, and archaeology,
among others. Teaching the principles and methods involved to several thousand stu-
dents and consulting for many industrial clients have further broadened Dr. Russ’ experi-
ence and the scope of applications for image processing and analysis.
After retirement, Dr. Russ was Research Director for Rank Taylor Hobson, a manu-
facturer of precision instrumentation. He continues to write, to consult for a variety
of companies (and to provide expert testimony in criminal and civil cases), to teach

xvii
© 2011 by Taylor & Francis Group, LLC
workshops worldwide on image processing and analysis, and to review publications and
funding proposals. He is active in the Microscopy Society of America, the Microbeam
Analysis Society, the Society of Photo-Optical Instrumentation Engineering (SPIE), the
International Society for Stereology, is a board member of the Society for Quantitative
Morphology, and a Fellow of the Royal Microscopical Society, and has presented invited
lectures and workshops for these and other organizations. On November 16, 2006, the
New York Microscopical Society awarded John Russ the Ernst Abbe Memorial Award for
his contributions to the field of microscopy as a developer of computer-assisted micros-
copy and image analysis.

xviii Author
© 2011 by Taylor & Francis Group, LLC
 1
Acquiring Images

Human reliance on images for information

H
umans are primarily visual creatures. Not all animals depend on their eyes, as we do,
for most of the information received about their surroundings (the characteristics of
human vision are discussed in Chapter 2). This bias in everyday life extends to how
we pursue more technical goals as well. Scientific instruments commonly produce images
to communicate their results to the operator, rather than generating audible tones or emit-
ting a smell. Space missions to other planets and equally arduous explorations of the ocean
depths always include cameras as major components, and the success of those missions is
often judged by the quality of the images returned. This suggests a few of the ways in which
humans have extended the range of natural vision. Optical devices such as microscopes and
telescopes allow us to see things that are vastly smaller or larger than we could otherwise.
Beyond the visible portion of the electromagnetic spectrum (a narrow range of wavelengths
between about 400 and 700 nanometers) there are sensors capable of detecting infrared and
ultraviolet light, X-rays, and radio waves, and perhaps soon even gravity waves. Figure 1.1
shows an example, an image presenting radio telescope data in the form of an image in which
color represents the Doppler shift in the radio signal. Such devices and presentations are used
to further extend imaging capability.
Signals other than electromagnetic radiation can be used to produce images, too. Novel new
types of microscopes that use atomic-scale probes to “feel” the specimen surface present their
data as images (Figure 1.2). The data so collected may represent the surface elevation and
topography, but other signals, such as the lateral drag force on the probe, may also be used.
Acoustic waves at low frequency produce sonar images, while at gigahertz frequencies the
acoustic microscope produces images with resolution similar to that of the light microscope,
but with image contrast that is produced by local variations in the attenuation and refraction
of sound waves rather than light. Figure 1.3 shows an acoustic microscope image of a sub-
surface defect, and Figure 1.4 shows a sonogram of a baby in the womb.
Some images such as holograms or electron diffraction patterns record brightness as a func-
tion of position, but are unfamiliar to the observer. Figure 1.5 shows an image of an electron
diffraction pattern from a transmission electron microscope, in which the atomic structure of
the samples is revealed (but only by measurement and to those who know how to interpret

1
© 2011 by Taylor & Francis Group, LLC
 Figure 1.1 Radio astronomy pro-
duces images such as this view of
Messier 33 (generated with data
from telescopes of the National
Radio Astronomy Observatory,
a National Science Foundation
Facility managed by Associated
Universities, Inc.). These are
often displayed with false colors
to emphasize subtle variations
in signal strength or - as in this
example - Doppler shift.

Figure 1.2 Atomic force micro-

scope image of human chromo-
somes (courtesy S, Thalhammer,
F. Jamitzky, Helmholtz Zentrum
München, Germany).

the data). Other kinds of data, including weather maps with specialized symbols, graphs of
business profit and expenses, and charts with axes representing time, family income, choles-
terol level, or even more obscure parameters, have become part of daily life, as illustrated in
Figure 1.6. The latest developments in computer interfaces and displays make extensive use
of graphics, to take advantage of the large bandwidth of the human visual pathway. Tufte
(1990, 1997, 2001) in particular has demonstrated the power of appropriate graphics to com-
municate complex information.
There are some important differences between human vision, the kind of information it
extracts from images, and the ways in which it seems to do so, as compared to the use of
imaging devices based on computers for scientific, technical, or forensic purposes. Humans

2 The Image Processing Handbook, Sixth Edition

© 2011 by Taylor & Francis Group, LLC
 (a)

(b)

Figure 1.3 Acoustic microscope image of voids in solder bond beneath a GaAs die: (a) die surface; (b)
acoustic image showing strong signal reflections (white areas) from the surface of the voids (courtesy
J. E. Semmens, Sonoscan Inc).

are especially poor at judging color or brightness of objects and features within images unless
they can be exactly compared by making them adjacent. Human vision is inherently com-
parative rather than quantitative, responding to the relative size, angle, or position of several
objects but unable to supply numeric measures unless one of the reference objects is a mea-
suring scale. Overington (1976; 1992) disagrees with this widely accepted and documented
conclusion but presents no compelling counter evidence. Chapter 2 illustrates some of the
consequences of the characteristics of human vision as they affect what is perceived.
This book’s purpose is not to study the human visual pathway directly, but the overview in
Chapter 2 can help the reader to understand how humans see things so that we become bet-
ter observers. Computer-based image processing and analysis use algorithms based on human
vision methods in some cases, but also employ other methods that seem not to have direct
counterparts in human vision. In particular, some image processing methods are based on the
physics of the image formation and detection process (Sharma, 2005).
Many of the examples and much of the analysis presented in this text involve images from
various types of microscopes. The three classes of imaging applications that generally offer

Acquiring Images 3
© 2011 by Taylor & Francis Group, LLC
Figure 1.4 Surface reconstruction of sonogram imaging, showing a 26 week old fetus in the womb.

Figure 1.5 A convergent beam electron diffraction (CBED) pattern from an oxide microcrystal, which
can be indexed and measured to provide high accuracy values for the atomic unit cell dimensions.

4 The Image Processing Handbook, Sixth Edition

© 2011 by Taylor & Francis Group, LLC
Figure 1.6 Typical graphics used to communicate news information include one-dimensional plots
such as stock market reports and two-dimensional presentations such as weather maps.

the most straightforward types of images are microscopy, aerial (and satellite) imagery, and
industrial quality control. That is because in those situations there is the greatest knowledge
and/or control over the imaging geometry and the illumination of the scene. In more general
“real world” cases the analysis and interpretation of the image contents can be much more
difficult. Objects may lie at various distances from the camera, which complicates determin-
ing size, may have different lighting, which alters their color, and may even partially obscure
other objects. Crime scene and accident photographs are often taken under difficult condi-
tions, from less than optimum points of view, and with variable lighting, so that their analysis
can be challenging.
The basic techniques for image processing and measurement are much the same for images
regardless of their source or scale. Images ranging from microscopy to astronomy, images
formed with light photons or sound waves, magnetic resonance or scanning profilometers,
have much in common and the techniques for dealing with their imperfections, enhancing
and extracting the details, and performing measurements utilize the same algorithms and
techniques, which are set out in the following chapters. The interpretation of the measure-
ments, as presented in later chapters, does require some specialization for different viewing
geometries, but is fundamentally independent of magnification.

Acquiring Images 5
© 2011 by Taylor & Francis Group, LLC
Video cameras
When the first edition of this book was published in 1990, the most common and affordable
way of acquiring images for computer processing was with a video camera. Mounted onto a
microscope or copystand, in a satellite or space probe, or using appropriate optics to view
an experiment, the camera sent an analog signal to a separate “frame grabber” or analog-to-
digital converter (ADC) interface board in the computer, which then stored numeric values in
memory (Inoué, 1986; Inoué & Spring, 1997).
The basic form of the original type of video camera is the vidicon, illustrated in Figure 1.7. It
functions by scanning a focused beam of electrons across a phosphor coating applied to the
inside of an evacuated glass tube. The light enters the camera through the front glass surface
(and a thin metallic anode layer) and creates free electrons in the phosphor. These vary the
local conductivity of the layer, so the amount of current that flows to the anode varies as the
beam is scanned, according to the local light intensity. This analog (continuously varying)
electrical signal is amplified and, as shown in Figure 1.8, conforms to standards of voltage
and timing (the standards and timing are slightly different in Europe than the US, but the basic
principles remain the same).
Digitizing the voltage is accomplished by sampling it and generating a comparison voltage.
The child’s game of “guess a number” illustrates that it takes only eight guesses to arrive at a

Deflection and Focusing Coils Anode

Grid

Electron Beam
Cathode Glass Tube

Phosphor
Coating

Figure 1.7 Functional diagram of a vidicon tube. Light striking the phosphor coating changes its local
resistance and hence the current that flows as the electron beam scans in a raster pattern.

0.7 volt range

0.3 volt sync pulse

52 sec. picture width

63.5 sec. horizontal scan interval

Figure 1.8 Standard RS-170 video signal shows the brightness variation along one scan line (ranging
between 0 volts = black and 0.7 volts = white).

6 The Image Processing Handbook, Sixth Edition

© 2011 by Taylor & Francis Group, LLC
 Figure 1.9 Example of an image show-
ing pincushion distortion, as well as
loss of focus and vignetting in the
edges and corners.

value that defines the voltage to one part in 256 (the most widely used type of ADC). The first
guess is 128, or half the voltage range. If this is (e.g.) too large, the second guess subtracts
64. Each successive approximation adds or subtracts a value half as large as the previous. In
eight steps, the final (smallest) adjustment is made. The result is a number that is conveniently
stored in the 8-bit memory of most modern computers.
The tube-type camera has several advantages and quite a few drawbacks. Scanning the beam
with electromagnetic or electrostatic fields can produce a distorted scan (pincushion or barrel
distortion, or more complicated situations) and is subject to degradation by stray fields from
wiring or instrumentation. Figure 1.9 shows an example of pincushion distortion, as well
as vignetting and loss of focus. Maintaining focus in the corners of the image takes special
circuitry, and the corners may also be darkened (vignetting) by the reduction in effective lens
aperture and the additional thickness of glass through which the light must pass. The sealed
vacuum systems tend to deteriorate with time, and the “getter” used to adsorb gas molecules
may flake and fall onto the phosphor if the camera is used in a vertical orientation. The
response of the camera (voltage vs. brightness) approximates the logarithmic response of film
and the human eye, but this varies for bright and dark scenes. Recovery from bright scenes
and bright spots is slow, and blooming can occur in which bright light produces spots that
spread laterally in the coating and appear larger than the features really are, with “comet tails”
in the scan direction.
There are, however, some advantages of the tube-type camera. The spatial resolution is very
high, limited only by the grain size of the phosphor and the size of the focused beam spot.
Also, the phosphor has a spectral response that can be made similar to that of the human eye,
which sees color from red (about 0.7 µm wavelength) to blue (about 0.4 µm). Adaptations of the
basic camera design with intermediate cathode layers or special coatings for intensification are
capable of acquiring images in very dim light (e.g., night scenes, fluorescence microscopy).

CCD cameras
The tube-type camera has now been largely supplanted by the solid-state chip camera, the
oldest and simplest form of which is the CCD (charge-coupled device). The camera chip con-
tains an array of diodes that function as light buckets. Light entering the semiconductor raises
electrons from the valence to the conduction band, so the number of electrons is a direct lin-
ear measure of the light intensity. The diodes are formed by photolithography, so they have a

Acquiring Images 7
© 2011 by Taylor & Francis Group, LLC
Figure 1.10 The basic principle of CCD operation, illustrated as a set of buckets and conveyors (after
Janesick, 2001).

perfectly regular pattern with no image distortion or sensitivity to the presence of stray fields.
The devices are also inexpensive and rugged compared to tube cameras. CCDs were first
invented and patented at Bell Labs in 1969 (George Smith was awarded the 2009 Nobel Prize
in Physics for this invention), and have now largely displaced film in consumer and profes-
sional still and movie cameras.
The basic operation of a CCD is illustrated in Figure 1.10. Each bucket represents one “pixel”
in the camera (this word has a lot of different meanings in different contexts, as explained
below, so it must be used with some care). With anywhere from a few hundred thousand to
several million detectors on the chip, it is impractical to run wires directly to each one in order
to read out the signal. Instead, the electrons that accumulate in each bucket due to incident
photons are transferred, one line at a time, to a readout row. On a clock signal, each column
of pixels shifts the charge by one location. This places the contents of the buckets into the
readout row, and that row is then shifted, one pixel at a time but much more rapidly, to dump
the electrons into an amplifier, which produces an analog voltage signal that may be sent out
directly or measured to produce the numeric output from a digital camera.
The simplest way of shifting the electrons is shown in Figure 1.11. Every set of three elec-
trodes on the surface of the device constitutes one pixel. By applying a voltage to two of the
electrodes, a field is set up in the semiconductor that acts like a bucket. Electrons are trapped
in the central region by the high fields on either side. Note that this does not reduce the area
sensitive to incoming photons, because electrons generated in the high field regions quickly
migrate to the low field bucket where they are held. By changing the voltage applied to the
regions in six steps or phases, as shown in the figure, the electrons are shifted by one pixel.
First one field region is lowered and the electrons spread into the larger volume. Then the
field on the other side is raised, and the electrons have been shifted by one-third of the pixel
height. Repeating the process acts like a conveyor belt and is the reason for the name “charge-
coupled device.”

8 The Image Processing Handbook, Sixth Edition

© 2011 by Taylor & Francis Group, LLC
 φ1
φ2
φ3

Insulator t1 t 2 t3 t4 t5 t6 t1
Semiconductor
1 pixel period

Electron
Wells
Transferring
Charge

Figure 1.11 Varying voltages on a set of three electrodes shifts electrons from one pixel to another in a
CCD.

One significant problem with the chip camera is its spectral response. Even if the chip is
reversed and thinned so that light enters from the side opposite the electrodes, very little blue
light penetrates into the semiconductor to produce electrons. On the other hand, infrared light
penetrates easily and these cameras have red and infrared (IR) sensitivity that far exceeds that
of human vision, usually requiring the installation of a blocking filter to exclude it (because
the IR light is not focused to the same plane as the visible light and thus produces blurred
or fogged images). Figure 1.12 shows this spectral response, which can be further tailored
and extended by using materials other than silicon. The chip can reach a high total efficiency
when antireflective coatings are applied, limited primarily by the “fill factor” — the area frac-
tion of the chip that contains active devices between the narrow ditches that maintain electri-
cal separation. Also, the chip camera has an output that is linearly proportional to the incident
light intensity, convenient for some measurement purposes but very different from human
vision, the vidicon, and photographic film, which are all approximately logarithmic.
Human vision notices brightness differences of a few percent, i.e., a constant ratio of change
rather than a constant increment. Film is characterized by a response to light exposure which
(after chemical development) produces a density vs. exposure curve such as that shown in
Figure 1.13. The low end of this curve represents the fog level of the film, the density that is
present even without exposure. At the high end, the film saturates to a maximum optical den-
sity, for instance based on the maximum physical density of silver particles or dye molecules.
In between, the curve is linear with a slope that represents the contrast of the film. A steep
slope corresponds to a high-contrast film that exhibits a large change in optical density with
a small change in exposure. Conversely, a low-contrast film has a broader latitude to record a
scene with a greater range of brightnesses. The slope of the curve is usually called “gamma.”
Many chip cameras include circuitry or processing that changes their output from linear to
logarithmic so that the image contrast is more familiar to viewers. The more expensive con-
sumer cameras and most professional cameras include the possibility to read the “raw” linear
data as well as the converted image.

Acquiring Images 9
© 2011 by Taylor & Francis Group, LLC
 Figure 1.12 Spectral response:
(a) Silicon based chip.
(b) Color sensors in the human eye,
              (a)
which are commonly identified as
red, green and blue sensitive but
cover a range of long, medium and
short wavelengths.

             (b)

When film is exposed directly to electrons, as in the transmission electron micrograph, rather
than photons (visible light or X-rays), the response curve is linear rather than logarithmic. Many
light photons are needed to completely expose a single silver halide particle for development,
but only a single electron is needed. Consequently, electron image films and plates are often
very high in density (values of optical density greater than 4, which means that 9999/10000 of
incident light is absorbed), which creates difficulties for many scanners and requires more than
8 bits to record.
The trend in camera chips has been to make them smaller and to increase the number of
pixels or diodes present. Some scientific cameras, such as that used in the Hubble telescope,
occupy an entire wafer. But for consumer devices, making each chip one-third, one-quarter,
or even two-tenths of an inch in overall (diagonal) dimension places many devices on a single
wafer and allows greater economic yield. It also requires smaller, less costly lenses. Putting
more pixels into this reduced chip area (for more spatial resolution, as discussed below)
makes the individual detectors small, but the ditches between then have to remain about the
same size to prevent electrons from diffusing laterally. The result is to reduce the total effi-
ciency markedly. Some devices place small lenses over the diodes to capture light that would

10 The Image Processing Handbook, Sixth Edition

© 2011 by Taylor & Francis Group, LLC
 Saturation

Figure 1.13 Response of photographic film. The

central portion of the curve shows a linear

Film Density
increase in density (defined as the base-ten
logarithm of the fraction of incident light that
is transmitted) with the logarithm of exposure. Linear Range
(Slope = Gamma)
High (“hard”) contrast corresponds to a steep
curve, while low (“soft”) contrast gives a less
steep curve and films have a greater dynamic
range. Fog
Reciprocity
Level
Failure

   Log (Exposure)

otherwise fall into the ditches, but these add cost and also are not so uniform as the diodes
themselves (which are typically within 1% across the entire chip).
The other, and more important, effect of making the detectors small is to reduce their capacity
for electrons, called the well capacity. A typical 15 µm pixel in a scientific grade CCD has a
capacity of about 500,000 electrons, which with low readout noise (as can be achieved in spe-
cial situations) of a few electrons gives a dynamic range greater than photographic film. Even
larger well capacity and dynamic range can be achieved by combining (binning) more detec-
tors for each stored pixel by using more steps in the phase shifting during readout. Reducing
the area of the detector reduces the well size, and with it the dynamic range.
Increasing the noise, for instance by reading out the signal at video rates (each horizontal line
in 52 µs for US standard definition video), dramatically reduces the dynamic range so that a
typical consumer grade video camera has no more than about 64 distinguishable brightness
levels (expensive studio cameras meet the broadcast video specification of 100 levels). Since
with the chip camera these are linear with brightness, they produce even fewer viewable gray
levels, as shown in Figure 1.14. This performance is much inferior to film, which can distin-
guish thousands of brightness levels.
CMOS (Complementary Metal-Oxide Semiconductor) chips can also be used as image sensors,
and in terms of sheer numbers are now more common than the original CCD devices. They
are primarily used in relatively inexpensive consumer cameras and camera phones, although
some have found their way into digital single lens reflex cameras. The conversion of light
photons to electrons functions in the same way as in the CCD chip. The differences start with
the way the signal is read out. In the CMOS designs there are from two to four transistors
immediately adjacent to the light sensor which convert the charge to a voltage and amplify the
signal. In principle, this means that any pixel in the array can be read out directly, addressing
a pixel by row and column just as in a memory chip (Figure 1.15). This is different from the
CCD method of “sweeping” the charge out to one corner of the array, reading all of the pixels
in a fixed order.
The space taken up by these control transistors reduces the “fill factor” or active area of the
chip that is sensitive to light, but this is often compensated for by placing lenses over each
detector to collect light from the dead areas and direct it to the active sensor. The lenses, and
the use of individual amplifiers for each pixel, generally make the sensors in a CMOS detec-
tor less uniform than those in the CCD array, producing a fixed pattern that can be compen-
sated for in software (requiring recording an image with uniform illumination). In addition to
the fixed pattern noise, the CMOS detectors usually have a greater amount of random noise

Acquiring Images 11
© 2011 by Taylor & Francis Group, LLC
 Figure 1.14 Comparison of visibility of
gray level steps from linear (equal
steps) and logarithmic (equal ratios)
detectors:
(a) Plots of intensity.
(b) D
 isplay of the values from (a).

superimposed on the image signal because of the separate amplifiers, additional wiring and
its associated capacitance and thermal noise, and greater dark current. The very small active
regions (at least in the smaller chips used in pocket cameras and phones, and particularly as
the pixel counts have risen to several million) have small well capacities, resulting in limited
dynamic range for the images. The images are usually stored with 8 bits per channel, because
of the way memory is traditionally organized, but often do not have that much actual bright-
ness resolution.
Larger area CMOS chips are also made which have larger detectors and consequently a greater
well capacity and greater dynamic range. One advantage of the CMOS designs as used in more
expensive cameras arises from the fact that the circuitry to access the pixels can be arranged
along two adjacent sides of the array (addressing the rows and columns, respectively). That
makes it possible to carefully trim away the chip on the other two sides, and arrange four of
the chips together to produce a larger sensor with higher pixel counts. This approach, com-
bined with the use of much larger sensors to achieve greater sensitivity and dynamic range,

12 The Image Processing Handbook, Sixth Edition

© 2011 by Taylor & Francis Group, LLC
 Figure 1.15 Schematic diagram of a typical
CMOS detector. Each active light sensor
(green) has additional transistors that are
connected to addressing and output lines.

has led some manufacturers to prefer CMOS detectors as large as a traditional film negative
for digital single-lens reflex cameras.
The advantages of CMOS sensors lie primarily in three areas: they consume much less power,
and so give better battery life; the amplifier and digitization circuitry can be placed on the
same chip to reduce size and perhaps increase ruggedness; and the production methods for
the wafers are essentially the same as those used for other standard silicon devices such as
memory and processors, whereas CCD wafers require unique processing. The latter advantage
is somewhat offset by the fact that high quality CMOS sensors do require somewhat custom-
ized fabrication for greater uniformity and fewer noise-producing imperfections than can be
tolerated in other devices. While the cost to fabricate CMOS sensors is less than for CCD, the
design costs are much higher. Of course, for devices that are to be produced in large quantity,
this is a minor factor. The overall trend has been for CMOS sensors to continue to improve
in quality and performance, and while the advantages of the CCD sensor are still important
for most technical applications, it is wise to consider the trade-offs on a case-by-case basis
(Nakamura, 2006; Holst & Lomheim, 2007).

Camera artifacts and limitations

There are several problems with video cameras using chips which contribute to the specific
types of defects present in the images that must be dealt with by subsequent processing. One
is the fact that many video signals are interlaced (Figure 1.16). With high-definition video,
and with digital still cameras, the image is scanned progressively. Interlacing is a clever trick
to minimize visual flicker in broadcast television images, accomplished with tube cameras by
scanning the electron beam in the same interlace pattern as the display television set. With a
chip camera, it requires that the array be read out twice for every 30th of a second frame, once
to collect the even numbered lines and again for the odd numbered lines. In fact, many cam-
eras combine two lines to get better sensitivity, averaging lines 1 and 2, 3 and 4, 5 and 6, and
so on, in one interlace field, and then 2 and 3, 4 and 5, 6 and 7, etc. in the other. This reduces

Acquiring Images 13
© 2011 by Taylor & Francis Group, LLC
 Figure 1.16
   (a) Interlace scan covers even
Line
1
numbered lines in one six-
2 tieth-second field, and even
3
4 numbered lines in a second
5
6 field.
7
8
9
(b) W hen motion is present
10 (either in the scene or caused
11
12 by camera motion), this pro-
13
14 duces an offset in the com-
Field 1 Field 2 plete image.
               (a)

(b)

vertical resolution but for casual viewing purposes is not noticeable. Motion can cause the
even and odd fields of a full frame to be offset from each other, producing a significant deg-
radation of the image, as shown in the figure. A similar effect occurs with stationary images
if the horizontal retrace signal is imprecise or difficult for the electronics to lock onto; this is
a particular problem with signals played back from consumer video tape recorders. (Moving
images are also distorted with progressive scan cameras, due to the time required to read from
the top of the image to the bottom.)
During the transfer and readout process, unless the camera is shuttered either mechanically
or electrically, photons continue to produce electrons in the chip. This produces a large

14 The Image Processing Handbook, Sixth Edition

© 2011 by Taylor & Francis Group, LLC
background signal that further degrades dynamic range and may produce blurring. Electronic
shuttering is usually done line-at-a-time so that moving images are distorted. Some designs
avoid shuttering problems by doubling the number of pixels, with half of them opaque to
incoming light. A single transfer shifts the electrons from the active detectors to the hidden
ones, from which they can be read out. Of course, this reduces the active area (fill factor) of
devices on the chip, reducing sensitivity by 50%.
The high speed of horizontal line readout can produce horizontal blurring of the signal, again
reducing image resolution. This is partially due to inadequate time for the electrons to diffuse
along with the shifting fields, to the time needed to recover electrons from traps (impurities
in the silicon lattice), and partially to the inadequate frequency response of the amplifier,
which is a trade-off with amplifier noise. Even though the individual electron transfers are
very efficient, better than 99.999% in most cases, the result of being passed through many
such transfers before being collected and amplified increases the noise. This varies from one
side of the chip to the other, and from the top to the bottom, and can be visually detected in
images if there is not a lot of other detail or motion to obscure it.
Many transfers of electrons from one detector to another occur during readout of a chip, and
this accounts for some of the noise in the signal. Purely statistical variations in the produc-
tion and collection of charge is a relatively smaller effect. The conversion of the tiny charge
to a voltage and its subsequent amplification is the greatest source of noise in most systems.
Readout and amplifier noise can be reduced by slowing the transfer process so that fewer elec-
trons are lost in the shifting process and the amplifier time constant can integrate out more of
the noise, producing a cleaner signal. Cooling the chip to about –40° also reduces the noise
from these sources and from dark current, or thermal electrons. Slow readout and cooling are
used only in non-video applications, of course. Digital still cameras use the same chip technol-
ogy (but much higher numbers of detectors) as solid state video cameras, and produce higher
quality images because of the slower readout. Janesick (2001) discusses the various sources
of noise and their control in scientific CCDs of the type used in astronomical imaging (where
they have almost entirely replaced film) and in space probes.

Color cameras
Color cameras can be designed in three principal ways, as shown in Figures 1.17, 1.18,
and 1.19. For stationary images (which includes many scientific applications such as
microscopy, but excludes “real-time” applications such as video), a single detector array
can be used to acquire three sequential exposures through red, green and blue filters,
respectively (Figure 1.17), which are then combined for viewing. The advantages of this
scheme include low cost and the ability to use different exposure times for the different
color bands, which can compensate for the poorer sensitivity of the silicon chip to short
wavelength (blue) light.
Many high-end consumer and most professional grade video cameras use three sensors
(Figure 1.18). A prism array splits the incoming light into red, green, and blue components,
which are recorded by three different sensors whose outputs are combined electronically to
produce a standard video image. This approach is more costly, since three chips are needed,
but for video applications they need not be of particularly high resolution (even a high-defini-
tion video camera has many fewer pixels than a digital still camera). The optics and hardware
to keep everything in alignment add some cost, and the depth of the prism optics makes it

Acquiring Images 15
© 2011 by Taylor & Francis Group, LLC
 Figure 1.17 Schematic diagram of a color
wheel camera with red, green and blue
filters. The fourth filter position is empty,
allowing the camera to be used as a mono-
chrome detector with greater sensitivity for
dim images (e.g., fluorescence microscopy).

  

Figure 1.18 Schematic diagram of the prisms

and dichroic filters for a three chip color
camera.

  

impractical to use short focal length (wide angle) lenses. This design is rarely used in digital
still cameras.
Video images are often digitized into a 640 × 480 array of stored pixels (the dimensions of
the VGA display that was once standard for personal computers), but this is not the actual
resolution of the image. The broadcast bandwidth limits the high frequencies and eliminates
any rapid variations in brightness and color. A standard definition video image has no more
than 330 actual elements of resolution in the horizontal direction for the brightness (lumi-
nance) signal, and about half that for the color (chrominance) information. Color information
is intentionally reduced in resolution because human vision is not very sensitive to blurring
of color beyond boundary lines.
Of course, video signals can be further degraded by poor equipment. Recording video on
consumer-grade tape machines can reduce the resolution by another 50% or more, particularly
if the record head is dirty or the tape has been used many times before (an unfortunately very

16 The Image Processing Handbook, Sixth Edition

© 2011 by Taylor & Francis Group, LLC
 Figure 1.19
   (a) Stripe and
(b) B
 ayer filter patterns
used in single chip
cameras.

(a)               (b)

common problem with forensic examination of surveillance video is that the tapes are played
— over and over — for visual examination by local police so that by the time professionals get
them, the oxide coating — and the information — has been damaged or even removed). Video
images are not very high resolution, although HDTV (high definition television) has improved
things somewhat. Consequently, video technology is usually a poor choice for scientific imag-
ing unless there is some special need to capture “real time” images (i.e., 25–30 frames per
second) to record changes or motion. Digital still cameras have largely replaced them, as they
produce much higher resolution images with greater dynamic range.
Most digital cameras use a single pixel array, often of very high pixel (detector) count, with a
color filter that allows red, green, and blue light to reach specific detectors. Different patterns
may be used (Figure 1.19), with the Bayer pattern being very common (invented by Kodak
researcher Bryce Bayer and the basis for U.S. Patent 3,971,065 “Color Imaging Array,” issued in
1976). Notice that it assigns twice as many detectors for green as for red or blue, which mim-
ics to some extent the human eye’s greater sensitivity to green. The problem with the single-
chip camera, of course, is that the image resolution in each color channel is reduced. The red
intensity at some locations must be interpolated from nearby sensors, for example. It is also
necessary to design the filters to give the same brightness sensitivity in each channel. If this is
not done well, a herring-bone pattern (often referred to as a “zipper”) appears in images of a
uniform gray test card and color fringes appear along contrast edges in the picture, as shown
in Figure 1.20.
Interpolation techniques for Bayer pattern color filters reduce the image resolution as com-
pared to the number of individual detectors in the camera (which is generally the speci-
fication advertised by the manufacturer). Inherently, this “demosaicking” process involves
trade-offs between image sharpness, details, noise, processing time and conversion artifacts.
The quality of the result, judged by its ability to preserve sharp boundaries in brightness
while minimizing the introduction of color artifacts, varies inversely with the computational

Acquiring Images 17
© 2011 by Taylor & Francis Group, LLC
Figure 1.20 Example of “zipper” patterns resulting from poor interpolation in a single-chip digital
camera.

requirements. A comparison of several patented methods can be found in Ramanath (2000)

and Shao et al. (2005), Tamburino et al. (2010), and Guanara et al. (2010). The combination
of the specific color filter array arrangement and the camera’s interpolation firmware leaves
a signature in images that can be used in some cases to identify the model of camera used to
photograph a scene, even to identify specific fixed pattern noise from an individual camera,
and to detect alterations made to the image later (Bayram et al., 2006; Swaminathan et al.,
2007; Farid, 2008).
Pattern noise is not unique to single-chip cameras with a color filter array. Three-chip cameras
also have potential problems because all chips have some slight variations in the output from
individual transistors. In a three-chip system these produce different variations in the red,
green, and blue output that increase the color variations in the images.
Another approach to color camera design, developed by Foveon Corp. and used in a few
cameras, creates three transistors at each pixel location, stacked on top of each other, using
CMOS technology. Blue light penetrates the shortest distance in silicon and is detected in the
topmost transistor. Green light penetrates to the second transistor and red light penetrates
to the bottom one. The output signals are combined to produce the color information. This
approach does not suffer from loss of spatial resolution due to interpolation, but has potential
problems with consistent or accurate color fidelity.

Camera resolution
The signal coming from the silicon detector is analog, even if the digitization takes place within
the camera housing or even on the same chip, so the interpolation is done in the amplifier
stage. In most cases, the actual image resolution with a single chip camera and filter arrange-
ment is one-half to two-thirds the value that might be expected from the advertised number
of pixels in the camera, because of this interpolation. And some cameras record images with
many more stored pixels than the chip resolution warrants in any case. Such interpolation and
empty magnification contribute no additional information in the image.

18 The Image Processing Handbook, Sixth Edition

© 2011 by Taylor & Francis Group, LLC
Comparing cameras based on actual resolution rather than the stated number of recorded
pixels can be difficult. It is important to consider the multiple meanings of the word “pixel.”
In some contexts, it refers to the number of light detectors in the camera (without regard to
any color filtering, and sometimes including ones around the edges that do not contribute
to the actual image but are used to measure dark current). In some contexts it describes
the number of recorded brightness or color values stored in the computer, although these
may represent empty magnification. In other situations it is used to describe the displayed
points of color on the computer monitor, even if the image is shown in a compressed or
enlarged size. It makes much more sense to separate these various meanings and to talk
about resolution elements when considering real image resolution. This refers to the num-
ber of discrete points across the image that can be distinguished from each other and is
sometimes specified in terms of the number of line pairs that can be resolved. This is one-
third to one-half the number of resolution elements, since at least one element is needed
for the line and one for the space between lines. It depends on the amount of brightness
contrast between the lines and the spaces, and the amount of noise (random variation)
present in the image.
The situation is even more complicated with some digital still cameras that shift the detec-
tor array to capture multiple samples of the image. The most common method is to use
a piezo device to offset the array by half the pixel spacing in the horizontal and vertical
directions, capturing four images that can be combined to more or less double the resolu-
tion of the image as data are acquired from the gaps between the original pixel positions.
For an array with colored filters, additional shifts can produce color images with resolu-
tion approaching that corresponding to the pixel spacing. Some studio cameras displace
the entire sensor array to different regions of the film plane to collect tiles that are sub-
sequently assembled into an image several times as large as the detector array. Of course,
the multiple exposures required with these methods means that more time is required to
acquire the image.
Rather than a two-dimensional array of detectors, it is also possible to use a linear array (or
sometimes three, one each with red, green, and blue filters) that is swept across the image
plane to acquire the data. This method is common in desk-top scanners (which for many
applications are perfectly usable image acquisition devices). It has also been used in studio
cameras, and some light microscopes accomplish the same thing by moving the stage and
specimen under the optics so that an image of an entire 1 × 3 inch slide can be obtained
with high spatial resolution. The image file produced is huge; special software is required
to efficiently access the stored array (Bacus & Bacus, 2000, 2002) and to interactively deliver
a selected portion of the image data as the user varies position and magnification. Network
access to such stored images also presents bandwidth challenges, but facilitates collaboration
and teaching.
With either a single-chip or three-chip camera, the blue channel is typically the noisiest
due to the low chip sensitivity to blue light and the consequent need for greater amplifi-
cation. In many cases, processing software that reduces image noise using averaging or
median filters (discussed in Chapter 4) can be applied separately to each color channel,
using different parameters according to the actual noise content, to best improve image
appearance.
Digital cameras using the same chip technology as a video camera can produce much better
image quality. This is due in part to the longer exposure times, which collect more electrons
and so reduce noise due to statistics and amplification. Also, the slower readout of the data
from the chip, which may take a second or more instead of 1/60th of a second, reduces

Acquiring Images 19
© 2011 by Taylor & Francis Group, LLC
readout noise. Digital still cameras read out the data in one single pass (progressive scan), not
with an interlace. By cooling the chip and amplifier circuitry to reduce dark currents, integra-
tion (long exposures up to tens of seconds, or for some astronomical applications many min-
utes) can be used to advantage because of the high dynamic range (large well size and large
number of bits in the digitizer) of some chip designs. In addition, the ability to use a physical
rather than electronic shutter simplifies chip circuitry and increases fill factor. The number of
pixels in video cameras need not be any greater than the resolution of the video signal, which,
as noted above, is rather poor. In a digital still camera, very high pixel counts can give rise to
extremely high resolution, which rivals film in some cases.
There is also an interesting cross-over occurring between high end consumer and professional
scientific grade cameras. In addition to dedicated cameras for attachment to microscopes or
other separate optics, manufacturers are producing consumer single-lens reflex cameras with
enough resolution (15 to 20 million pixels at this writing) that it is becoming practical to use
them in technical applications, and simple optical attachments make it easy to connect them
to microscopes or other instruments (and of course the camera may also be removed and used
for other purposes). The camera may be tethered directly to a computer, but in many cases it
is more practical to record the images to memory chips that are later downloaded to the com-
puter. Professional digital cameras with large, high resolution detector arrays, interchangeable
lenses, etc., are providing capabilities that compete with traditional 35mm and larger film
cameras. Every manufacturer of cameras has recognized the shift away from film and toward
digital recording, and an incredibly wide variety of cameras is now available, with new devel-
opments appearing frequently.
The benefits of a camera with a large number of sensors (high pixel count), as well as large
individual sensors (large well size and consequent high dynamic range), seem obvious and
desirable. For some applications, high pixel counts are not so important. At high optical
magnification, the important limitation is the optical resolution. In the rather typical setup
of my bench microscope, with a 10x (low magnification) objective lens, the image projected
onto the chip by the transfer optics covers about 1600 µm width on the specimen. With a
100× (high magnification) objective lens that becomes 160 µm. For a camera with 3600 ×
2400 sensors (less than 10 megapixels) the low magnification image is recorded at about 1
pixel per micron, adequate for the resolution of the optics. The high magnification image
is recorded with 90 pixels per micron. Since the optical resolution of the microscope under
optimum conditions is about 0.5 µm with the 100× lens, this produces a vast and unneces-
sary oversampling. At low magnifications, or for viewing fine detail in large scenes (such as
aerial and satellite imagery), high pixel counts make sense. When the limitation on resolu-
tion lies with the optics, it may not.

Focusing
Regardless of what type of camera is employed to acquire images, it is important to focus the
optics correctly to capture the fine details in the image. Often the human eye is used to per-
form this task manually. In some situations, such as automated microscopy of pathology slides
or surveillance tracking of vehicles, automatic focusing is required. This brings computer
processing into the initial step of image capture. Sometimes, in the interests of speed, the
processing is performed in dedicated hardware circuits attached to the camera. But in many
cases the algorithms are the same as might be applied in the computer (described in Chapter

20 The Image Processing Handbook, Sixth Edition

© 2011 by Taylor & Francis Group, LLC
5), and the focusing is accomplished in software by stepping the optics through a range of
settings and choosing the one that gives the “best” picture.
Several different approaches to automatic focus are used. Cameras used for macroscopic
scenes may employ methods that use some distance measuring technology, e.g., using high
frequency sound or infrared light, to determine the distance to the subject so that the lens
position can be adjusted. In microscopy applications this is impractical, and the variation
with focus adjustment captured in the image itself must be used. Various algorithms are used
to detect the quality of image sharpness, and all are successful for the majority of images in
which there is good contrast and fine detail present. Each approach selects some implementa-
tion of a high-pass filter output which can be realized in various ways, using either hardware
or software, but must take into account the effect of high frequency noise in the image and
the optical transfer function of the optics (Green et al., 1985; Firestone et al., 1991; Boddeke
et al., 1994; Sun et al., 2004; Buena-Ibarra, 2005; Bueno et al., 2005; Brazdilova & Kozubek,
2009; Shim et al., 2010.)

Electronics and bandwidth limitations

Video cameras of either the solid-state chip or tube type produce analog voltage signals cor-
responding to the brightness at different points in the image. In the standard definition RS-170
signal convention, the voltage varies over a 0.7-volt range from minimum to maximum bright-
ness, as shown above in Figure 1.8. The scan is nominally 525 lines per full frame, with two
interlaced 1/60th-second fields combining to make an entire image. Only about 480 of the
scan lines are usable, with the remainder lost during vertical retrace. In a typical broadcast
television picture, more of these lines are lost due to overscanning, leaving about 400 lines in
the actual viewed area. The time duration of each scan line is 62.5 µs, part of which is used
for horizontal retrace. This leaves 52 µs for the image data, which must be subdivided into
the horizontal spacing of discernible pixels. For PAL (European) television, these values are
slightly different based on a 1/25th-second frame time and more scan lines, and the resulting
resolution is slightly higher.
Until recently in the United States, broadcast television stations were given only a 4-MHz
bandwidth for their signals, which must carry color and sound information as well as the
brightness signal. This narrow bandwidth limits the number of separate voltage values that
can be distinguished along each scan line to a maximum of 330, as mentioned above, and this
value is reduced if the signal is degraded by the electronics or by recording using standard
videotape recorders. Consumer-quality videotape recorders reduce the effective resolution
substantially; in “freeze frame” playback, they display only one of the two interlaced fields,
so that only about 200 lines are resolved vertically. Using such equipment as part of an image
analysis system makes choices of cameras or digitizer cards on the basis of resolution (actually
the number of sampled pixels) irrelevant.
There is a major difference between the interlace scan used in standard definition television
and a non-interlaced or “progressive” scan. The latter gives better quality because there
are no line-to-line alignment or shift problems. Most high definition television (HDTV)
modes use progressive scan. The format requires a higher rate of repetition of frames to
fool the human eye into seeing continuous motion without flicker, but it has many other
advantages. These include simpler logic to read data from the camera (which may be incor-
porated directly on the chip), more opportunity for data compression because of redun-
dancies between successive lines, and simpler display and storage devices. Practically all

Acquiring Images 21
© 2011 by Taylor & Francis Group, LLC
Exploring the Variety of Random
Documents with Different Content
 THE FLORA OF HANCOCK. 63 Sambucus Canadensis
(common elder). Sanibueus pubens (red-berried elder). MADDKK
FAMILY. lialium asprellum (i-ough bcdstraw). Galium trifldum (small
bedstraw). Galium apariue (goose-grass). Galium triflorum (sweet-
scented bedstraw). Mitchella prepens (partridge berry).
Ceplialanthus occidentalis (button-bush). Houstouia ca;rulea
(bluets). COMPOSITE FA.MILV. Cirsium lanceolatum (common
thistle). Cirsium arvense (Canada thistle). Cirsium discolor (two-
colored thistle). Cirsium pumilum (pasture thistle). Cirsium muticum
(swamp thistle). Onorpordon acanthium (cotton thi.stle).
Xanthumstrumarium(cocklebur thistle). Ambrosia artemisia^folia
(Roman wormwood). Panacetum vulgare (common tansy). Artemisia
absinthium (common wormwood). Krechthites hieracifolia
(fireweed). Gnaphalium polycephalum (everlasting) . Guaphalium
uliginosum (low cudweed). Antennaria margaritacea (pearly
everlasting). Antennaria plantaginifolia (plantainleaved everlasting).
Eupatorium purpureum (Joe-tye weed). Eupatorium perfoliatum
(thoroughwort) . TussiJage farl'au (coltsfoot.) senecio aureus (golden
ragwort). Inula Helenium (elecampagne). solidago bi-color (golden
rod). Solidago latifolia. >-olidago arguta. ■^olidago altissima.
Solidago Canadensis. solidago gigantea. ^olidago rigida. Aster
corymbosus. Vster macrophyllus. Aster cordifolius (star wort). Aster
patens. Aster Xovic Anglia:. Aster puniceus. Aster longifolius. Aster
multiflorus (frostweed). Aster tradescanti. Aster umbellatus. Erigeron
Philadelphicum (tleabaue). Erigeronbellidit'olium (robins' plantain).
Erigeron strigosum (daisy fleabane). Erigeron annuum (larger daisy
flea bane). Erigeron Canadensc (horseweed). Achillea millefolium
(yarrow). Maruta cotula (Mayweed). Chrysanthemum leneanthemum
vulgare (whiteweed— ox-eye dai.sy). Helenium autumnale (sneeze-
weed). Bidens frondosa (beggar-ticks). IJidens connata (swamp
beggar-ticks). Bidens chi-ysanthemoides (larger beggar-ticks)
Ileliopsis lasvis. Rcndbeckia hirta. Hendbeckia laciniata (common
cone flower). (Jichorum intybus (succory or chiccory). Leontodon
autumnale (fall damlelion orhawkbet). Hieracium Canadense
(hawkweed). Nabalus altissimus (tall white lettuce). Nabalus albus
(common white lettuce). Nabalus fraseri (lions'-foi't) . Taraxicum
dens-leonis (dandelion). LOBELIA FAMILY. Lobelia cardinalis (cardinal
flower). Lobelia inflata (Indian tobacco). Lobelia kalmii. Lobelia
dortmanna (water lobelia). CAMPANULA FAMILY. Campanula apai-
inoides (marsh bellwort. HEATH FAMILV. Gaylussacia rcsinosa
(huckleberry). Gaylussacia frondosa (danglebury). Vaccinium
Pennsylvanicum (early dwart blueberry). Vaccinium Canadense (high
blueberry). Vaccinium corymbo.sum (high swamp blueberry).
Vaccinium o.xycoccus (small cranljcrry). Vaccinium macrocarpon
(large cranberry). Gaultheria procumbens (checkerberry). Cassandra
calyculata (leather-leaf). Andromeda polifolia. Andromeda ligustrina.
Kalmia latifolia (mountain laurel). Kalmia angustifolia Hambkill, or
sheep laurel). Rhododendron maximum (rliododendron). Azalea
nudiflora (election pinks). Clethra alnifolia (white alder).
 64 HISTORY OF HANCOCK, NEW HAMPSHIRE. CMogenes
hispedula (creeping snowberrj-). Pyrola rotundifolia. Pyrola elliptica.
Pyrola cblorantha (wintergreeii). Pyrola secunda. Cbimapbila
umbellata fpipsissewa). Monotropa uniflora (Indian pipe). Monotropa
bypopitys (false beecbdrops). HOLLT FAJULY. Ilex verticillata (black
alder) . Xemopantbis Canadensis (mountain boUy). PLANTAIN
FAMILY. Plantago major (plantain). Plantago lanceolata (ribgrass, or
Euglisb plantain). PRIMROSE FAMILY. Trientalis Americana (star
flower). Lysimacbia tbyi-siflora (loose strife). Lysimachia stricta.
Lysimacbia quadrifolia. Lysimacbia eiliata. Anagallis arvensis (poor
man's motber glass). BLADDERAVORT FAMILY. Utricularia inflata
(bladderwort). Utricularia vulgaris. BROOM-RAPE FAMILY. Epipbegus
Virginiana (beecb-drops). FIGWORT FAMILY. Verbascum tbapsus
(mullein). Veronica Americana. Veronica ofiicinalis (speedwell).
Linnaria Canadensis (toad flax). Linnaria vulgaris (butter and eggs).
Gerardia tenuifolia. Gerardia pedicularia. Mimulus ringens (monkej-
flower). Hysantbes gratioloides (false pimpernel. Gratiola Virginiana (
b edge-by ssop). Cbelone glabra (snake-bead). Pedicularis
Canadensis (lousewort). Melampyrum America num (cow wbeat).
VERVAIN FAMILY. Verbena bastata (blue vervain). Verbena urticifolia
(wbite ver.vain). MINT FAMILY. Tencrium Canadense (germander).
Tricbostema dicbotomum (blue curls). Isantbus cserulens (false
pennyroyal). Mentba viridis (spearmint). Mentba piperita
(peppermint). Mentba Canadensis (wild mint) . Lj-copus Virginicus
(bugleweed). Hedeoma pulegioides (American pennyroyal).
Collinsonia Canadensis (borse balm). Calamintba clinopodium (basil).
Monarda flstulosa (wild bergamot;-. Xepeta cataria (catnip). Xepeta
gleeboma (gi-ound ivy). Bi-uuella vulgaris (self-beal). Scutellaria
galei'iculata (skullcap). Galeopsis tetrabip (bemp nettle;. Leonurus
cardiaca (motberwort). Stacbys palustris (bedge nettle). BORAGE
FAMILY. Echium vulgare (viper's bugloss). Myosotis palusti-is var laxa
(wild forgetme-not). Ecbinospermum lappula (stickseed).
Cynoglossum morisoni (beggar's lice). Lycopsis arvensis (small
bugloss). CONVOL\TJLUS FAMILY. Calystegia sepium (bedge
bindweed). NIGHTSHADE FAMILY. Solanum dulcamara (bittersweet).
Nicandra pbysaloides (apple of Pei-u;. Datura sti-amonium (tborn
apple). GENTIAN FAMILY. Gentiana crinita (ft-inged gentian).
Gentiana Andrewsii (closed gentian). Gentiana sapouaria (soapwort
gentian). Menyantbes trifoliata (buckbean). Limnantbemum
lacunosum (floating beart). DOGBANE FAMILY. Apocynum
androssemifolium (dogbane). Apocynum cannabinum (Indian bemp).
MILKWEED F.tMILY. Asclepias incarnata (swamp milkweed).
Asclepias cornuti (common milkweed). Asclepias pbytolaccoides
(poke milkweed). Asclepias verticillata (wborled milkweed). OLI>-E
FAMILY. Fraxinus Americana (wbite asb). Fraxinus pubescens (red
asb). Fraxinus sambucifolia (black asb). BIRTmVORT FAMILY.
Aristolocbia serpen t.aria (Virginia. snakeroot). i
 THE FLORA OF HANCOCK. 65 AMAUAMII FAMII.V.
Aniarantus panieulatus. AmaranUis rctroflexus (pijrwceci).
AuKirantus alhus. in IKWIIKAT lAMILV. Polygonum aviculare (goose-
grass, or doorwooil). Polygoiiuin I'ersicaria (lady's thumb).
Polygonum aiuphibium (wator Pi-rslcaI'olygonuni acre (water
smartwoed). Polygonum Hydropiper. Polygonum arifolium (tear-
thumb). Polygonum sagittatum. Polygonum oonvolvulus (black
bindweed). I'olygonum dumetorum (climbing buckwheat). IJumex
orbiculatus (great water dock). Rumex Britaunica (pale dock).
I!unie.x crispus (common dock}. Rumex sanguineus ( b 1 o o d y ■ v
e i n e d dock). Rumex Acetosella (sheep-sorrel). ilEZEKEUM FAMILY.
Dirca palustris (leatherwood). NKTTLE F.A.MILV. Ulmus Americana
(Auierican, or white elm). Urtica dioica (nettle). I'rtica gracilis (fence-
nettle). I'rtica urens (small nettle). Taportea Canadensis (wood-
nettle). I-LAXE-TKEE FAMILY. Plataiuis occidentalis (buttonwood).
^VAL.^■L■T FAMILY. •luglans cinerea (butternut). ( arya alba (shag-
bark hickory). . OAK FAMILY. Quercus alba (white oak). Quercus
rubra (red oak). Fagus ferruginea (American beech). Coi-ylus
Americana (American hazel). Corylus rostrata (beaked hazel). Osteya
Virginica (American liop-hornbeam). Carpinus Americana
(hornbeam). SWEET GALE FAMILY. -Myrica gale (sweet gale).
(Jomptonia asplenitolia (sweet fern). lilKClI FAMILY. ftctula lenta
(sweet, or black birch). Betula lutea (yellow birch). Betula alba, var.
populitolia (American white birch). Betula papyracea (paper, or
canoe birch). .\lnus incana (alder). \VILLOW FAMILV. Salix cordata
(willow). Salix longifolia. Salix alba. Salix lucid.a. Populus tremuloides
(American aspen). Populus grandidentata. Populus dilatata
(Lombardy poplar). Populus balsamifua var. candicans (balm of
(Jilead). PINE FAMILY. Pinus rigida (pitch pine). Pinus resinosa (red,
or Norway pine). Pinus strobus (white pine). Abies nigra (black
spruce). Abies Canadensis (hemlock spruce). .4.bies balsamea
(balsam-flr). Larix Americana (American larch). Taxus baccata, var.
Canadensis (ground hemlock). ARUM FA.MILY. Arisiema triphyllum
(Indian turnip). Calla palustris (water arum, or wild calla).
Symplocarpus foetidus (skunk cabbage). A corns (jalamus (sweet
flag). C-\X-TAIL FAMILY. Typlia latifolia (cat-tail flag). Typlia
angu.stifolia (nari'ow-leaved flag). A V A T E K I" L A N T A I N I' A
M I L \ . Trigliicliin i);i lustre (arrow-grass). .Schciich/i ria alisnia
(water plantain). Sagitlaiiii sariabilis (arrow-head). I'IClvEKEL WEED
FAMILY. Pon(h;deria cordata (pickerel weed). ORCHIS FAMILi .
Orchis spectabilis (showy orchis). Habenaria flmbriata (large
purplefringed orchis). Habenaria psycodes (smaller p ur pie fringed
orchis). Habenaria lacera(ragged-friuged orchis). Habenaria
blephariglottis (white-fringed orchis).
 66 HISTORY OF HANCOCK, NEW HAMPSHIRE. Habenaria
virescens. Habenaria hyperborea. ' Spirantbes gracilis (ladies'
tresses). Gordyera pubescens (rattlesnake plantain) . Pogonia
opbioglossoides. Arethusa bulbosa. Cypripedium acaule (lady's
slipper). Micro'stylis monophyllos. IRIS FAMILY. Iris Virginica (slender
blue flag). Iris versicolor (lai-ge blue flag). Sisyrincbium Bermudiana
(blue-eyed grass) . LILY FAMILY. Trillium erectum (birchroot, or
purple trillium) . Trillium erythrocarpum (painted trillium). Medeola
virginica (Indian cucumber root) . Veratnim viride (Indian poke) .
Uvularia sessilifolia (bellwort). Clintonia borealis. Streptopus rosens
(twisted stalk) . Smllacina bifolia (false Solomon's seal). Smilaeina
racemosa. Polygonatum biflorum (small Solomon's seal).
Polygonatum giganteum (large Solomon's seal). Lilium
Pbiladelphicum (wild orangered lily). Lilium Canadense (Canada lily).
Erythronium Americanum (adder's tongue). PIPE WORT FAMIIA'.
Eriocaulon septangulare (pipewort.) FERX FAMILY'. Polypodium
vulgai-e (polypody). Adiantum pedatum (maiden-hair). Pteris
aquilina (common brake). Woodwardia virginica (chain fern).
Asplenium trichomeus. Asplenium ebeneum. Asplenium
angustifolium. Asplenium thelypleroides (spleenwort). Asplenium
felix-foemina (lady fern). Phegopteris polypodioides (beech fern).
Phegopteris dryopteris (beech fern). Aspidium Thelypteris. Aspidium
cristatum (wood fern) . Aspidium spinulosum (shield fern). Aspitlium
marginale. Aspidium acrostichoides. Cystopteris fragilis. Onoclea
sensibilis (sensitive fern). Woodsia obtusa. Dicksonia punctilobula.
Osmunda cinnamonea (cinnamon fern). Osmunda regalis (royal
fern). Botrychium ternatum. Botrychium virginicum (moonwort).
Opheoglossum v u 1 g a t u ni (adder's tongue). CLUB MOSS
FAMILY. Lycopodium lucidulum. Lycopodium annotinum. Lycopodium
dendroideum. Lycopodium cleratum. Lycopodium complanatum.
Selaginella rupestris. These tables have been prepared with much
care, and are exhaustive. We will only give in addition, in this place,
the location of some of the different kinds of trees found in the
original forest. In the south-east part of the town much of the
original forestgrowth was white pine. Many of these trees would
probably have been reserved for "His Majesty's Royal Navy," as was
the case in towns earlier settled, had not "His Majesty" just about
that time been compelled to relinquish all claim to the soil and its
productions. The stump of a tree, in good preservation, is standing
on land now owned by Augustus W. Gray, in Bennington, the timber
of which was used in the construction of the first bridge over the
Contoocook, at the point where an iron bridge was recently built,
near Goodell's works.
The text on this page is estimated to be only 17.54%
accurate

67 iJuous trees jf^n and on eties), black sugar-maple ^

part of the h was very Ve been the ay, who was through the ^\^y,.
-. ■''\ ising of the •ked on the ' r.\Ct^~Mr^^ Nottingr \': . \ 'v 1/ /|d
(No. 1 on JU.^'^''-^ ../-seriate monu;^p5:^'l\'ii.il-W* m any of ■,tV
-^wn, he had that time. ^/ ^ ; o our patrons, ■5,. o such an ex',
injr houses in iiicf tl);ii rime luliincil house licrs ill
subse'j'uAA^tuntorest. Tlic S . , ,,s the origina. .jt; ^ifli tlicri' uia\•;:-
^^vritcs : " I toft ■•>L/^;ioorihuu-ociv Jf^^v occui)y tli(!ir
The text on this page is estimated to be only 28.13%
accurate

EARLY SETTLERS. 67 Towards the west, with an increased

altitude, deciduous trees became more numerous. Near the center of
the town and on towards the west, were found beech, birch (several
varieties), black and white ash, and white maple trees. The rock or
sugar-maple was the prevailing growth in some sections of the west
part of the town; the red oak was also frequently found. Black ash
was very plenty near Tenney's pond, and this was said to have been
the cause of the settlement in that locality of Daniel Tenney, who
was a basket-maker. Hemlock and spruce were scattered through
the town then, as now, and near the center there was an abundance
of Norway pine, whence the names given to the pond, plain, and hill.
A tract of land bordering on Moose brook, known as the bog, was
covered with a dense growth of cedar, which was of great value to
the early settlers. This timber was killed by the rising of the water,
caused by Jesse Rogers' dam, at the spot marked on the map XII.
CHAPTER V. EARLY SETTLERS. John Grimes, the first settler of
Hancock, came from Nottingham ^yest, by the way of Peterboro', in
the summer of 1764. He built his log cabin (to which he brought his
family in the spring of 1765) not far from the south shore of Half-
moon pond (No. 1 on the map^). The citizens of Hancock erected
an appropriate monument on this S])Ot in 1884. As his name does
not appear in any of the papers connected with the incorporation of
the town, he had probably returned to Nottingham West previous to
that time. ' The map ol" Hancock, which we take great pleasure in
presentinfj to our patrons, is drawn from one jjublished in lMo8, with
corrections and additions to sucli an extent tliat it is substantially a
new map. The names of thost; occui)yin;r houses in l.SoJS arc;
retained in their p'ropcr place. Ilousis i reeled mi new siles >in(:e
thai lime have been ailded, with the iiai.i.- .il their ..ei-up.'inl- in -
eiipi. A baii.lnned house spots are nunibei-cd. As there will !..■
eMiilinnal allu-ion 1m llitei-'<. it="" has="" not="" t.eeii=""
deeme="" iieee-.ar="" lo="" ilie="" iiaiii.="" in="" this=""
.="">ainloTie,l niill -ite- are iniin l.eiv, I \^ ill, i.'oniaii niiiiieraN -. ■ .-
liaplei- on •-ourees of Income. .\( Mil ion- lia\e al-o licen made, -ix
in- I la- locali..n- ol tlie two rail r. liilN. name- o| hill-, >lream-, and
lake-, and sarion- points ol iiilere-t. The courses (jI the streams lia\e
al-o liei n eluniKed in some instances, as the origiua. nuip was
somewhat imiieileet in I hi- ic'ard. Till- artist. Mis- l.elia .1. \\
 68 HISTORY OF HANCOCK, NEW HAMPSHIRE. Some of his
children subsequently returned to Hancock, and his widow died
there in 1827. Moses Mokrisox came from Peterboro' as early as
1769. He settled on what is known as Morrison hill, east of Half-
moon pond. William Lakin came from Grotou, Mass., probably in
1770. He settled on the farm now owned and occupied by his great-
grandson, Joshua S. Lakin. Geobge McClouky came from Peterboro'.
He settled on the farm now owned and occupied by Willis A.
Washburn. The date of his settlement is not known, but probably it
was as early as 1775. Samuel Mitchell, Jr., came from Peterboro'. He
was a resident of the town in 1776. John Young and John Espie were
also here at that time, but nothing more is known of them. John
Moore came from Bedford as early as 1773, He settled on the north-
east side of Miller's mountain (No. 57). Six children were born here;
but about the year 1790 he sold his farm to Robert Mathews and
returned to Bedford. Thomas English was a brother-in-law of John
Moore, and came from Bedford in 1776. Settled near Mr. Moore (No.
56), and removed to Antrim in 1779. John Miller came from
Goffstown as early as 1775. Place of settlement. No. 55. Thomas
Miller's name appears as one of the signers of the memorial relative
to the incorporation of Antrim, in 1776. He was probably a brother of
John Miller. Jonathan Bennett, Stephen Bennett, and Asa Adams
came from Shirley, Mass. They settled on and near what is now
known as the "Old Stoddard road," on the north-east side of Mount
Skatutahkee. The fii-st town-meeting was held at the house of
Jonathan Bennett. At that time it was supposed to be the center of
the town. Jonathan Barnet was at Bunker Hill from Society Land.
hmrts, minds, and hands. I have suggested the farm products by
vegetables and grain arranged in one corner, by giapes and^pples in
another corner; the raising of animals and the dairy business by
horses' heads, sheep, and cattle in another division of the border
[portraits of these faithful friends] ; some few manufactures by
specimens of cutlery and farming tools; saw- and grist-mills by logs,
lumber, and bags of grain; and the "literary and religions interests by
books, papers, and cards, with the names of subjects." Thus we
have here illustrated the religious, literary, and artistic tastes of the
people, as well as the diflerent industries in which they have been
and still are actively engaged. The design is a great addition to the
value of our work ; but perhaps the greatest interest attached to it is
tlie fact that it is the production of one of our own artists; thus
giving in itself an illustration of Hancock's dioicest treasures — its
gifted men and women.
 EARLY SETTLERS. 69 He also signed the memorial relative
to the incorporation of Antrim in 1776. It \% probable that both
names apply to the same person, but not certain. Stephen Bennett
was in town as early as 1778, and Asa Adams as early as 1779.
RoBEKT DuxcAX camc from Londonderry to Society Land in 1774.
Pie settled at first on the east side of the river. In the early part of
1779 he removed to what is now Hancock, and settled on Norway
hill. A little later he removed to the place where C. E. L. Hayward
now resides. William Williams and WiLLiAAr Williaais, Jk., came from
Shirley, Mass., locating at the place marked No. 24. Joel Russell
settled at the place marked No. 8G. Joseph Stmojs^ds came from
Groton, Mass., through New Ipswich, in the early part of 1779,
possibly in the fall of 1778. He settled on Norway hill. Oliver
Lawrence, John Cummings, John Cummings, Jr., and Reuben
Cummings, came from Hollis. They were, without doubt, here at the
time of the petition for the incorporation of the town ; had pui-
chased their farms, and were engaged in clearing their land, but did
not remove their families here till a little later. Mr. Lawrence settled
on what is known as the Nathaniel Dow farm, in 1782. John
Cummings and John Cummings, Jr., at the place marked No. 88, at
about the same time. Reuben Cummings came many years later, if
at all. Timothy Moors came from Groton, Mass., thi#^igh New
Ipswich, probably in the early part of 1779. He settled on the place
marked No. 107. James Davis came from Paxton, Mass. He settled at
the place marked No. 30. David Ames came from Peterboro'. He
settled on a spot about three rods north-east of the Center station.
John Foster came from Deerfield, Mass. He settled on the place
marked " Mrs. Cummings." Arthur Graham came from Windham. He
settled at No. 46. Nathan Merriam is supposed to have boarded with
Stephen Bennett. Adam I/atterson settled a little to the north-east of
Jack's pond (No. 39). This makes up the list of those whose names
appear on different documents previous to the incorporation of the
town. Others
 70 HISTORY OF HANCOCK, NEW HAMPSHIRE. came before
the charter was granted, and within a few years quite a number
were added to the list of settlers, with whose history we are more
interested than we are with some of those whose names are here
enumerated, but want of space prevents their appearance in this
connection. The reader is referred to the genealogical record for
further information in this direction. The first settlers of Hancock
were subject to all the hardships incident to their position. It will be
noticed that their homes were far apart, and further on, as we trace
the successive settlements, we shall discover that they seemed
anxious to have all the room possible. Without any system of roads,
all communication must have been attended with a good deal of
difficulty. Their homes were of the rudest description, log cabins,
without floors or windows, and many of them without chimneys,
were all that at first they aspired to possess. The first glass window
in town was brought by Mrs. Moses Dennis from Andover, in 1784.
William Lakin made his home at first in an abandoned hunter's
camp. Many of the deserted housespots, of which there are so many
in town, were occupied but a few years, while many others of which
no mention is made, soon gave place to better homes near where
the first ones stood. We need not hei-e repeat the story of their
hardships, it is so familiar to all. Incidentally, however, we may be
able to refer to it in subsequent chapters. There werd^orae
compensations for the difficulties that they met. The virgin soil was
more fertile than it is at the present day ; as has been mentioned,
an abundance of fish were to be found in the streams and lakes,
which must have been a great help to them in the absence of other
food. In the lakes were to be found pickerel, perch, hornpouts or
catfish, and other fish. In the brooks were the delicious brook-trout
and suckers, while salmon were caught in the Contoocook. Wild
animals Avere also numerous. The larger and more dangerous, the
wolf, catamount, wild-cat, bear, deer, and beaver, have disappeared ;
while the fox, rabbit, woodchuck, hedge-hog, raccoon, mink, musk-
rat, otter, and a variety of squirrels are still to be found. Of birds we
would mention the eagle, hawk, crow, owl, blackbird, partridge,
robin, bobolink, bluejay, and sparrow. Bears were not uncommon
visitors of the sheep-pastures and pigstyes of the early settlers.
Joseph Symonds, Sr., caught a large one
 EAH1.\ SI'TI'I KKS. 71 in a trap soon after his arrival in
town, near Norway pond, which was to him a real acquisition, as the
flesh was said to have been excellent food. Mrs. James Duncan
drove a bear away from her pigs with a club soon after her
settlement here. Abraham Moore, who lived on the jklace marked
No. 83, on one occasion having started to carry a spinning wheel-
head to a neighbor's, met a bear on his way, which offered him the
usual challenge of a lorestle by the back. Mr. Moore declined the
challenge, and sent the wheel-head rolling toward him. The bear, not
comprehending this mode of warfare, beat a retreat, but Mr. Moore
did not care to follow to investigate, so as to ascertain where he
might be found. Other visits were made by the bears to the early
settlers, the last being made to Josiah Stone, Sr., not far from the
close of the Eighteenth century. Mr. Stone then resided on the farm
where Jonathan Bennett first settled. He had lost some of his lambs
and pigs. He mistrusted that a bear had been the thief, but was not
sure; it might have been foxes. He took his gun, with a moderate
charge of shot in it, and his dog, and went to the sheep pasture. He
met a bear, and discharged his gun, but only wounded and enraged
the bear. He ran, the bear after him. Just as he attempted to get
over a large log the bear caught hold of him and tore a part of his
clothing off. The dog then made an assault upon the bear from
behind, which diverted his attention to such an extent that Mr, Stone
escaped. It is needless to add that that dog was kept by the family
until he died of old age. Wolves were also numerous and
troublesome. Capt. John Cummings and his brother-in-law, Joseph
Symonds, were compelled to swing fire-brands, to keep them away
from their camp, nearly all of the first night of their stay in town.
They had some fresh meat in their camp, which the wolves were
determined to get. A hunter and trapper, who occupied for a time
the place marked Xo. 8, left his gun outside the door of his hut one
night, and was obliged to swing fire-brands till daylight, to keep at
bay a i)anther, which had got between him and his gun. Rachel
Adams (afterwards Mrs. Isaac Davis), whose home at that time was
with her brother, Asa Adams, made a visit one day with the
daughters of Moses Morrison, and was escorted to her home by
wolves.
 72 HISTORY OF HANCOCK, NEW HAMPSHIRE. Zebedee
Whittemore, who lived at No. 92, was treed one night by wolves,
and compelled to remain in that unpleasant position till morning,
although near his home. During the dark day of May, 1780, Mrs.
James Duncan was alone, her husband having gone to Society Land
for goods. The floor of her house was full of cracks. Looking through
them she saw the glaring eyes of a wolf. How long he remained
there we do not know, but it would be hard to realize a more
thrilling experience than that of this young wife and mother. The day
itself was one that would appal the stoutest heart. "There fell Over
the hloom and sweet life of the spring, Over the fresli earth and the
heaven of noon, A horror of great darkness, like the night In day, of
which the Norland sagas tell — The twilight of the gods. The low-
hung skyWas black with ominous clouds, save where its rim Was
fringed with a dull glow like that which climbs The crater's sides from
the red hell below. Birds ceased to sing, and all the barn-yard fowls
Koosted; the cattle at the pasture bars Lowed and looked
homeward; bats, on leathern wings. Flitted abroad; the sounds of
labor died; Men prayed, and women wept." Mrs. Duncan was alone
in the savage wilderness, her only companions two young children,
— one two years and a half old, and one barely two months. We can
only say of her as Whittier writes of Abraham Davenport, that she is
" A witness to the ages as they pass. That simple duty hath no place
for fear. " Mrs. Bezaleel Spaulding, on one occasion, went after her
cow, and lingering too long, to pick berries, Avas greeted by the
howls of wolves. Nothing daunted, she seized the cow by the tail
and made quick time for the barn. There is at Marshfield, Vt., a bell
which was once worn by sheep to keep off the \volves in what is
now the streets of Bennington. Deer, though not numerous, were
sometimes shot by the settlers. The last one shot in town was killed
by Isaac Fitch in 1818, near Antrim line. Mr. Fitch, who was a great
hunter, also shot the last lynx seen in town, at a somewhat later
date. Wild turkeys were frequently seen. Joseph Symonds, Sr., saw
one or more in the road near his house. Edmund Davis shot a large
 EARLY SETTLEKS. 73 oue near his home (No. 28). They
were frequently heard by the Goodhue family, who lived near Antrim
line. Their head-quarters seem to have been in what was then
known as "Tophet Swamp" (now called swamp woods), near the line
between Peterboro' and Hancock. Simeon Lakin, Sr., who had
become somewhat acquainted with their habits when a soldier,
determined to secure some of them. They were pretty quiet during
the day, but Mr. Lakin found their roosting-place, and scattered
unthreshed rye from this place to the cleared land near Samuel
Ames' mill (No. I). Here he built a bough house. After the turkeys
were well baited, Mr. Lakin and his neighbor, Nathan Brooks, who
lived at No. 101, went down in the night and waited for them to
come out for their morning meal. With the early dawn they made
their appearance, led by a staid, old gobbler. Mr. Brooks was so
excited that he fired at them without taking good aim. The ball
lodged in a tree; but all the injury inflicted on the turkeys was, they
were badly frightened! Mr. Lakin afterwards was able to capture
some of them in a trap. Wild pigeons %vere numerous, and large
numbers were captured in nets or "pigeon stands." Within forty
years, Eemington Tuttle was quite successful in capturing them.
Perhajjs in this connection it would not be out of place to give Moses
Morrisoirs celebrated • moose story. It is substantially as follows : —
"One morning Brother Duncan [they were brothers-in-law] came up
to our house and wanted to borrow some meat. I told Brother
Duncan I would be glad to let him have some, but could not, as we
were short at our house, too. I told him that there was a moose up
in the bog, and we Avould go right up and try and get it. He
consented and took his gun, and I took my 'little dog' and gun, and
we went to the bog. "We soon found the moose and shot it, and
began to skin and dress it. Brother Duncan was awful hungry^ and
began to cut off from one quarter ;ind to cook and eat it. The first I
knew he had eaten it all up ! When the moose was all dressed, I told
Brother Duncan to take his gun and one quarter, and start. I swung
the other two quarters, with the hide and my gun, on to my back,
and we started. Pretty soon Brother Duncan began to feel a little
sick, and I told him to throw his gun and the other quarter on the
top of my load. By-and-by the 'little dog ' got tired, and Brother
Duncan
 74 HISTORY OF HANCOCK, NEW HAMPSHIRE. put him on
the top of the load. Before we got out of the bog Brother Duncan
got so sick he thought he could go no farther, I was afraid wolves
might get him, so I told him to get on to a stump, and when I went
by, to get on. He did so, and I carried Brother Duncan, moose, 'little
dog,' and guns to hard ground," When we remember that a moose
will dress several hundred pounds, we can understand that it was
very kind in Mr. Morrison to be so helpful. Moose brook takes its
name from this story, which has not probably lost any thing in the
hundred years since it was first told. Other towns have pre-empted
many of Mr. Morrison's stories, so it may not be well to repeat them
here. We must, however, protest against the imputation that he was
" a mere story-teller," and that "it is for this only that his memory
has survived him." He was really the first permanent settler of the
town. The records show that he bore an honorable part in its early
history ; and although the historian of Peterboro' knew " of no one
who could furnish a record of his descendants," the statement
reveals the fact that that historian had a limited knowledge of the
history of at least otie adjoining town. CHAPTER VI. SOURCES OF
INCOME. The pioneers of Hancock had but few sources of income,
and stern necessity compelled them to practise the severest
economy. Of modern improvements and modern conveniencies they
knew nothing. Money was scarce, and most of the limited business
of the time was carried on by barter. Within the memory of the
present generation this method of doing business was common. Hay,
grain, and other crops, and heavy loads of all kinds, were moved in
summer as well as winter on sleds, for want of any thing better. One
of the early improvements was a vehicle unknown at the present
day, and hard to describe. It was simply a slender crotched tree,
with prongs some seven or eight feet long, to drag on
 SOURCES OF INCOME. 75 the ground ; the trunk serving
tlie juirpose of a sled-spire. On the two prongs was placed a box,
holding eight or ten bushels, called a car. Salmon Wood owned one
of these "cars." Amos Tenney was more progressive ; he sawed off
two sections of a large elm for wheels, and so constructed a rude
vehicle that at the time was considered a great advance on any thing
that had yet been used. It was a custom once a year to make formal
visits, using oxsleds to convey the family, with chairs and coverlets
for the convenience of the women and children. On these occasions
liquors were always served, and there was no little rivalry as to who
should be able to serve the best. The first chaise in town was owned
by John Clark, the second by Samjtson Tuttle, Sr. The first "pleasure
wagon " was owned by Prescott Hayward. For many items of
interest, in this connection, see centennial addresses, especially that
of Mr. Robinson. FORESTS. Lumber was plenty and of extra quality.
Unfortunately it was not well cared for. The forests were looked
ui)on as the natural enemies of the settlers here as elsewhere. A
considerable income was derived from the burning of the raonarchs
of the forest, and converting the ashes into potash. Andrew Seaton,
and perhaps others, carried on the business of making potash. His
place of business was where afterward Oliver Whhcomb lived.
Charcoal burning was also carried on to some extent during the first
sixty years of the town's history. Considerable lumber for building
purjioses has been exported from time to time. Timber has been,
and is, sold from the farms every year for the manufacture of
various kinds of wooden ware. The selling of wood for fuel has also
for many years been carried on to some extent, not only for home
consumption, but also to some of the neighboring villages, and since
our suj)erior railroad facilities, to towns and cities more remote.
Maple sugar has been made from the sap of the sugar- or rockmaple
every year since the first settlement of the town, and no doubt by
the Indians before the arrival of our ancestors. The early facilities for
this purpose were poor compared with
 76 HISTORY OF HANCOCK, NEW HAMPSHIRE. those
deemed necessary now. Instead of tapping the trees with an auger
or bit, it was the custom to "box" them with an axe, an operation
that in a few years would destroy the life of the tree. Rude wooden
troughs for catching and holding the sap, and kettles hung on poles
for boiling, have gradually given place to the nicely painted wooden
and tin buckets, and the improved boiling pans and evaporators of
to-day. Hancock maple sugar has been sent as far south as Georgia,
and as far west as California in our own country, and exported to
China. In quality it has no superior. Among the largest producers at
the pi-esent time, we will mention David Wilds, A. S. Wood, Milan E.
Davis, Albert Jaquith & Son, Charles Hayward & Son, Lakin & Tarbell,
Orland Eaton, and Ebenezer Ware ; some of whom have
manufactured from one to two tons in a season. Among others who
manufacture less in quantity, but of fine quality, we would mention
Asa D, Wood, John P. Hills, and Charles M. Sheldon'. CULTIVATED
CROPS. The first crop taken from the virgin soil was usually rye ;
considerable flaxseed was also sown. Rye and flaxseed were used in
matters of barter, indeed as a kind of currency. Taxes were often
paid with rye, also debts. Contracts were made with rye at so much
per bushel as a basis. From the flax was manufactured, in the homes
of the settlers, a large proportion of the lighter clothing that was
worn, to say nothing of the table linen, etc., used in the household.
It was expected that every man would have his clearing and ryefield
each year. The method of clearing the land where there was no
demand for the timber was to cut down the trees, "lop" the
branches, and after the wood was partially dried, to burn it on the
ground. The charred logs were then rolled together into huge piles,
and new fires were kindled and continued till they were entirely
consumed. It was customary for the neighbors to gather together
and make what they called " bees " to do this heavy work ; hence
the term " log-rolling," so often used in our own time. In order to
raise rye it was not necessary to plow the land ; in fact much of the
land of Hancock never could be plowed ; but the
 SOURCES OK INCOME. ( 7 seed was sown broadcast on the
burnt land and " scratched in " by hand. This process did not cease
with the early inhabitants, as the writer distinctly remembers having
had a "hand'' in tliat kind of work. Maize or Indian corn has always
been a staple crop in the town. There was a time when it was
exported to a certain extent, but more western corn is now imported
and consumed in town than that of home production. From the
beginning our farmers have given considerable attention to the
culture of potatoes, but not until recently have many been exported.
Formerly they were consumed at home, large quantities being fed
out to the stock. Wheat enough for home consumption was raised
for a time. The same may be said of oats and barley; but little grain
of any kind is now sown in the town. Of pease and beans enough
have been generally raised for home use, and for a long period "
bean porridge " Avas a staple article of food. The culture of flax was
long ago abandoned here. I doubt if many of those living ever saw
flax growing, or would recognize a field of flax were they to see it.
Grass is and always has been an important crop in Hancock. For a
few years after the land was cleared abundant crops were gathered.
The rougher portions of the land were gradually given up to
pasturage, and no better pastures could be found anywhere. One
who has long since passed away told me, some years ago, that he
sold a pair of oxen from his pasture in June, and the purchaser who
drove them to the Boston market, informed him afterwards that
there was not a field that he passed in the entire distance, into
which he would prefer to turn the oxen, than the pasture from
whence they were taken. The gradual failure of these permanent
pastures is perhaps the most discouraging feature in the agricultural
prospects of the town. Where sixty years ago could be found, in the
summer-time, an abundance of herds' grass, honey-suckle, and
clover, to-day there is only a scanty crop of "June grass,"^ although
in some jjlaces fair pasturage can yet be found. In the near future
the only remedy for a large proportion of the rougher land will be to
let it return to its original condition for a season. It is no more than
justice, however, to say that large tracts of the smoother land ' "
Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade

Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.

Let us accompany you on the journey of exploring knowledge and

personal growth!

ebookname.com