EEE 6512
Image Processing and Computer Vision
Lecture 2
Introduction to Digital Image Processing
Dr. Dapeng Oliver Wu
University of Florida
Department of Electrical and Computer
Engineering
Fall 2015
Lecture Outline
Applications of image processing
Image formation and perception
Image representation
Matlab primer
Image processing demos
Application Areas of Image Processing
Purpose of image processing
Improvement of pictorial information for human
interpretation
Processing of image data for storage, transmission,
and representation for autonomous machine
perception
Typical application areas
Television Signal Processing
Satellite Image Processing
Medical Image Processing
Computer Vision and Robot Control
Visual Communications
Biometrics for Law Enforcement
3
Television Signal Processing
Image brightness, contrast, color hue
adjustment
Television image enhancement
High Definition TV (HDTV)
Satellite Image Processing
Remote sensing
Climate
Satellite
Doppler radar
Geology
Land resource
Flood monitor
Doppler refers to the principle the
Austrian scientist Christian Doppler
discovered in 1842.
New York (from Landast-5 TM)
5
Medical Image Processing
Images are acquired to get information about
Anatomy and Physiology of a patient
Ultra Sound (US)
Magnetic Resonance Imaging
Positron Emission Tomography (PET)
Computer Tomography (CT)
XRays
6
Computed Tomography (CT)
Allan M. Cormack and Godfrey N. Hounsfield
won Nobel Prize in Physiology or Medicine in
1979 for development of CT
How CT was invented?
Cormack proposed the back-projection
method to generate a three-dimensional
image of the internals of a human body
from a large series of two-dimensional Xray images taken around a single axis of
rotation.
Hounsfield constructed the first CT system
practicable in medical care.
7
Magnetic Resonance Imaging (MRI)
Paul C Lauterbur and Peter Mansfield
won Nobel Prize in Physiology or
Medicine in 2003 for development of MRI
How MRI was invented?
Paul Lauterbur proposed a method that uses
gradient magnets along with the main magnet
to detect signals for creating 2D images.
Peter Mansfield proposed a signal processing
method to transform the detected signals into
a 2D image.
8
Computer Vision and Robot Control
Industry robots on production line
Unmanned operations
Autonomous Vehicle driving
DARPA Grand Challenge Race, LA to Vegas
AUVSI and ONR's Autonomous Underwater
Vehicle Competition (U.F. SubjuGator won the first
place in 2005, 2006, 2007)
International Micro Air vehicle Competition (U.F.
won every year since 1999)
Mars Rover
Visual Communication
Videophone
Tele-conferencing
Tele-shopping
Video coding
Video transmission
10
Biometrics for Law Enforcement
Biometric identification technology
Fingerprint
Face recognition
Iris
Retina
Palm
DNA
AR Face Database
11
Components in Digital Image Processing
Output are images
Wavelets
Waveletsand
and
Multiresolution
Multiresolution
processing
processing
Compression
Compression
Image
Image
restoration
restoration
Image
Image
enhancement
enhancement
Input
Image
Image
Image
acquisition
acquisition
Morphological
Morphological
processing
processing
Segmentation
Segmentation
Knowledge
Knowledgebase
base
Representation
Representation
&&description
description
Object
Object
recognition
recognition
12
Output are image attributes
Color
Colorimage
image
processing
processing
Image Formation (Image Capture)
Light source (: wavelength of the source)
E ( x, y, z , 1 , 1 , ) : insident light on a point, where (x, y, z) is world coordinate
of the point, (1 , 1 ) is the incident angle.
Each point of the scene has a reflectivity function.
r ( x, y, z , 2 , 2 , ) : reflectivity function, where (2 , 2 ) is viewing angle.
Light reflects from a point and the reflected light is captured by an
imaging device.
c( x, y, z , 2 , 2 , ) E ( x, y, z, 1 , 1 , ) r ( x, y, z , 2 , 2 , ) : reflected light
E ( x, y, z , 1 , 1 , )
c( x, y, z, 2 , 2 , )
Camera (c( x, y, z, 2 , 2 , ))
13
What is Light?
Light is a form of electromagnetic (EM) radiation, with
from 430nm to 790nm (visible light band).
14
Illuminating and Reflecting Light
Illuminating light sources:
emit light (e.g. the sun, light bulb, TV monitors)
perceived color depends on the emitted freq.
follows additive rule
R+G+B=White
Reflecting light sources:
reflect an incoming light (e.g. the color dye, matte
surface, cloth)
perceived color depends on reflected freq (=emitted
freq-absorbed freq.)
follows subtractive rule
C+M+Y=Black
15
Projection
Projection (P) from world coordinates (x, y, z) to
camera or image coordinates (x, y)
c p ( x, y, ) P (c( x, y, z , 2 , 2 , )
Two types of projections (P) of interest to us:
1. Perspective Projection
Objects closer to the capture device appear bigger. Most
image formation situations can be considered to be under
this category, including images taken by camera and the
human eye.
2. Orthographic Projection
This is unnatural. Objects appear the same size
regardless of their distance to the capture device.
16
Perspective Projection
Film
Cover
Object 1
Object 2
z1
z2
Pinhole
1 2 , z1 z 2
2 1
17
Orthographic Projection
Film
Object 1
Object 2
1 2
z1
1 2 , z1 z 2
z2
2 1
18
Math Involved in Image Formation
Euclidean geometry
Projective geometry
Formalize one of the central principles of
perspective art: parallel lines meet at infinity
and therefore are to be drawn that way
Affine geometry
Do not involve any notions of origin, length or
angle, but with the notion of subtraction of
points giving a vector.
19
Projective Geometry & Affine Geometry
20
Light Sensitivity
Each capture device has its sensitivity
function V().
V()
V()
The result is an image function which
determines the amount of reflected light
that is captured at the camera coordinates
(x, y). f ( x' , y ' ) c ( x' , y ' , )V ( )d
21
Human Eye System
Eye Anatomy
22
Receptors in the Retina
Rods
Night vision
Low acuity (low resolution)
Achromatic
Cones
Day vision
High acuity
Chromatic
Three sets
700nm (R), 546nm (G), 435nm (B)
23
Image Representation
Analog image -> Digital image
Sampling
Quantization
2D array of picture elements (pixels)
24
Grayscale Image
The intensity value of each pixel is 0 ~ 255
Matrix representation
160 162 166 154
162 158 122 69
60 55 79 94
58 55 99 109
25
Color Image
Three components
M = {R, G, B}
73 87
66 98
31 61
R , G , B
27 17
36 13
36 14
26
A Brief Matlab Tutorial
An interactive program from the
MathWorks for high-performance numeric
computation and visualization.
Refer to Matlab Primer for general use
Type help image to see functions in
image processing toolbox
Run demos in image processing toolbox
http://www.mathworks.com/products/image/demos.html
27
Simple Matlab Commands
>> x = imread('baboon.bmp');
>> whos
Name
Size
Bytes Class
x
256x256x3
196608 uint8 array
Grand total is 196608 elements using 196608 bytes
>> imshow(x);
>> g = rgb2gray(x);
>> imshow(g);
>> g(1:2, 1:2)
ans =
64
77
46
75
28
Image Processing Demos
Simple point processing
Special effects
Noise reduction
Image enhancement
Image restoration
Face detection
Image segmentation
29
Digital Negative
Diagonal flip
Horizontal flip
Original image
Simple point processing
30
Swirl
Wave
Rotation
Original Image
Special Effects
31
Noise Reduction
Degraded Image
Noise reduced Image
32
Image Enhancement
Observed Image
Enhanced Image
33
Image Restoration
Degraded Image
Restored Image
34
Face Detection
Face tracking in a video
Face recognition
35
Image Segmentation
Segmentation of different object in the
scene
36
High Performance Imaging Using Large Camera Arrays
Image mosaic: stitch multiple images into a
panoramic view
High speed video capture: convert space into time
E.g., 52 cameras, each with 30 fps, provide 1560
fps
Slowly replay high speed event (e.g., a moving
bullet through balloons). Can a human see a
moving bullet? Why not?
37
Synthetic Aperture Photography
Bennett Wilburn et al., High performance imaging
using large camera arrays, ACM SIGGRAPH 2005
38
Sprite
Sprite (complete background)
Video object (foreground)
Warping based on camera
motion, e.g., perspective,
bilinear, or affine mapping
Blending
A sprite, also referred to as mosaic, is an image composed of pixels
belonging to a video object visible throughout a video sequence.
39
Example of Mosaic
Merge 10 pictures into one picture
(how to remove the overlapped part?)
40
Homework 1
Use a digital camera or web cam to take a color picture of
yourself and save the picture as a JPEG file with filename
YourLastNameColor.jpg. Then use rgb2gray function in
Matlab to convert your RGB color picture to a gray-scale
image and save the gray-scale image as a JPEG file with
filename YourLastNameGray.jpg. (You will need to use
YourLastNameColor.jpg and YourLastNameGray.jpg for
image processing in other homework assignments.)
Submit the above two JPEG files through E-Learning web
site under the directory of Homework 1.
Due: 4pm, 9/9. Late submission will not be accepted.
You can also check the calendar at https://lss.at.ufl.edu/ for
all the assignment due dates.
41
Homework 1 (no submission for the
following items)
1. Read the Matlab Primer and get familiar with
Matlab. The Matlab primer can be found at
http://math.ucsd.edu/%7Edriver/21d-s99/matlab-primer.html
2. Type help image to see functions in image
processing toolbox
3. Run demos in image processing toolbox
http://www.mathworks.com/products/image/demos.html
42
Reading Assignment
Digital Image Processing, Chapter 1,
Chapter 2 (Section 2.1 2.3)
43
