Histogram Processing

 Histogram Equalization

 Histogram Matching

 Local Histogram Processing

 Using Histogram Statistics for Image

Enhancement

03/25/25 1
 Histogram Processing
Graphical Representation of Data
Associated with Digital Image
Frequency of occurrence of various gray levels
Histogram h( rk ) nk
rk is the k th intensity value
nk is the number of pixels in the image with intensity rk
K = 0, 1, ………..L-1
nk
Normalized histogram p( rk ) 
MN
nk : the number of pixels in the image of
size M N with intensity rk

03/25/25 Probability of occurrence of gray level rk 2

 Histogram Equalization

Normalized histogram is a probability density function (PDF) :

p(rk) = h(rk) / n = nk / n
- probability of occurrence of intensity level rk, where n is the total
number of pixels.
Equalized histogram is a cumulative distribution function (CDP):

Discrete values:
k
sk T (rk ) ( L  1) pr (rj )
j 0
k nj L 1 k
( L  1)   nj k=0,1,..., L-1
j 0 MN MN j 0
03/25/25 3
Histogram Equalization
The intensity levels in an image may be viewed as
random variables in the interval [0, L-1].
Let pr (r ) and ps ( s ) denote the probability density
function (PDF) of random variables r and s.

03/25/25 4
Histogram equalization Histogram
Linearisation requires
construction of a transformation
function sk

5
Example: Histogram Equalization
Suppose that a 3-bit image (L=8) of size 64 × 64 pixels (MN
= 4096) has the intensity distribution shown in following
table.

Get the histogram equalization transformation function and give

the ps(sk) for each sk.

03/25/25 6
Example: Histogram Equalization

0
s0 T (r0 ) 7 pr (rj ) 7 0.19 1.33 1
j 0
1
s1 T (r1 ) 7 pr (rj ) 7 (0.19  0.25) 3.08 3
j 0
s2 4.55  5 s3 5.67  6
s4 6.23  6 s5 6.65  7
s6 6.86  7 s7 7.00  7
03/25/25 7
Example: Histogram Equalization

03/25/25 8
03/25/25 9
03/25/25 10
 Digital Image With Various Gray Levels

Occurrence
6 6 7 7 6 0-0
5 2 2 3 4 1-0
2-3
3 3 4 4 5 3-4
5 7 3 6 2 4-4
7 6 5 5 4 5-5
6-5
7-4

5
4 4 5 4
3

0 1 2
3 4 5 6 7

Used to manipulate contrast & Brightness

Quality enhanced due to normalizing & histogram offers flat profile
11
 Image Dynamic Range, Brightness and Control
The dynamic range of an image is the exact subset of gray values ( 0,1,2, L-1) that are
present in the image. The image histogram gives a clear indication on its dynamic range.
When the dynamic range of the image is concentrated on the lower side of the gray
scale, the image will be dark image.
When the dynamic range of an image is biased towards the high side of the gray scale,
the image will be bright or light image
An image with a low contrast has a dynamic range that will be narrow and concentrated
to the
middle of the gray scale. The images will have dull or washed out look.
When the dynamic range of the significantly broad, the image will have a
imageand
contrast is the distribution of pixels high iform (Normalized to Flat profile).
will be ne ar un

12
 Histogram Components
Concentrated on the low side
of the gray scale

Biased toward the high side

Centered toward the middle

•Almost Uniform distribution

•Cover a broad range of gray scal
03/25/25
•Toward the equalization 13
14
Histogram Equalization
s T (r ) 0 r L  1
a. T(r) is a strictly monotonically increasing function
in the interval 0 r L -1;
b. 0 T (r ) L -1 for 0 r L -1.

T (r ) is continuous and differentiable.

ps ( s )ds  pr (r )dr
03/25/25 15
Histogram Equalization
r
s T (r ) ( L  1)  pr ( w)dw
0

ds dT (r ) d  r 
dr

dr
( L  1)

dr  
0
pr ( w) dw

( L  1) pr (r )

pr (r )dr pr (r ) pr (r ) 1
ps ( s )    
ds  ds  ( L  1) pr (r )  L  1
 
 dr 
03/25/25 16
Example
Suppose that the (continuous) intensity values
in an image have the PDF

 2r
 2
, for 0 r L-1
pr (r )  ( L  1)
 0, otherwise


Find the transformation function for equalizing

the image histogram.

03/25/25 17
Example
r
s T (r ) ( L  1)  pr ( w)dw
0

r 2w
( L  1)  2
dw
0 ( L  1)

2
r

L 1

03/25/25 18
Histogram Equalization
Continuous case:
r
s T (r ) ( L  1)  pr ( w)dw
0

Discrete values:
k
sk T (rk ) ( L  1) pr (rj )
j 0
k nj L 1 k
( L  1)   nj k=0,1,..., L-1
j 0 MN MN j 0
03/25/25 19
Question
Is histogram equalization always good?

No

03/25/25 20
Histogram Matching
Histogram matching (histogram specification)
—A processed image has a specified histogram
Let pr ( r ) and pz ( z ) denote the continous probability
density functions of the variables r and z. pz ( z ) is the
specified probability density function.
Let s be the random variable with the probability
r
s T ( r ) ( L  1)  pr ( w) dw
0

Obtain a transformation function G

z
G ( z ) ( L  1)  pz (t ) dt s
0

03/25/25 21
Histogram Matching
r
s T (r ) ( L  1)  pr ( w)dw
0
z
G ( z ) ( L  1)  pz (t )dt s
0

1
z G ( s ) G 1
T (r ) 

03/25/25 22
Histogram Matching: Procedure
 Obtain pr(r) from the input image and then obtain the
values of s r
s ( L  1)  pr ( w)dw
0

 Use the specified PDF and obtain the transformation

function G(z)
z
G ( z ) ( L  1)  pz (t )dt s
0

 Mapping from s to z

z G  1 ( s )
03/25/25 23
Histogram Matching: Example
Assuming continuous intensity values, suppose that an
image has the intensity PDF
 2r
 2
, for 0 r L -1
pr (r )  ( L  1)
 0, otherwise

Find the transformation function that will produce an
image whose intensity PDF is
 3z 2
 , for 0  z ( L -1)
pz ( z )  ( L  1) 3

 0, otherwise

03/25/25 24
 Histogram Matching: Example
Find the histogram equalization transformation for the input
image
2
r 2w r r
s T (r ) ( L  1)  pr ( w)dw ( L  1)  dw 
0 ( L  1) 2
0
L 1
Find the histogram equalization transformation for the specified
histogram
z 3t 2 z z3
G ( z ) ( L  1)  pz (t )dt ( L  1)  dt  s
0 0 ( L  1) 3 ( L  1) 2

The transformation function

1/3
1/3  2 r
2
 2 1/3
z  ( L  1) s 
2
 ( L  1)   ( L  1)r 
 L  1 
03/25/25 25
Histogram Matching: Discrete Cases
 Obtain pr(rj) from the input image and then obtain the
values of sk, round the value to the integer range [0, L-1].
k
( L  1) k
sk T (rk ) ( L  1) pr (rj )   nj
j 0 MN j 0
 Use the specified PDF and obtain the transformation
function G(zq), round the value to the integer range [0, L-
1].
q
G ( zq ) ( L  1) pz ( zi ) sk
i 0
 Mapping from sk to zq

1
zq G ( sk )
03/25/25 26
Example: Histogram Matching
Suppose that a 3-bit image (L=8) of size 64 × 64 pixels (MN
= 4096) has the intensity distribution shown in the following
table (on the left). Get the histogram transformation
function and make the output image with the specified
histogram, listed in the table on the right.

03/25/25 27
Example: Histogram Matching
Obtain the scaled histogram-equalized values,

s0 1, s1 3, s2 5, s3 6, s4 7,

s5 7, s6 7, s7 7.
Compute all the values of the transformation function G,
0
G ( z0 ) 7 pz ( z j ) 0.00 0
j 0

G ( z1 ) 0.00 0 G ( z2 ) 0.00  0
G ( z3 ) 1.05 1 G ( z4 ) 2.45  2
G ( z5 ) 4.55  5 G ( z6 ) 5.95  6
G ( z7 ) 7.00 7

03/25/25 28
Example: Histogram Matching

03/25/25 29
Example: Histogram Matching
Obtain the scaled histogram-equalized values,

s0 1, s1 3, s2 5, s3 6, s4 7,

s5 7, s6 7, s7 7.
Compute all the values of the transformation function G,
0
G ( z0 ) 7 pz ( z j ) 0.00 0
j 0

G ( z1 ) 0.00 0 G ( z2 ) 0.00  0
G ( z3 ) 1.05  1 s0 G ( z4 ) 2.45  2 s1
G ( z5 ) 4.55  5 s2 G ( z6 ) 5.95  6 s3
G ( z7 ) 7.00 7 s4 s5 s6 s7

03/25/25 30
Example: Histogram Matching
s0 1, s1 3, s2 5, s3 6, s4 7,
s5 7, s6 7, s7 7.

rk
0
1
2
3
4
5
03/25/25 6 31

7
Example: Histogram Matching
rk  zq
0 3
1 4
2 5
3 6
4 7
5 7
6 7
7 7
03/25/25 32
Example: Histogram Matching

03/25/25 33
Example: Histogram Matching

03/25/25 34
03/25/25 35
Local Histogram Processing
Define a neighborhood and move its center from
pixel to pixel

At each location, the histogram of the points in the

neighborhood is computed. Either histogram
equalization or histogram specification
transformation function is obtained

Map the intensity of the pixel centered in the

neighborhood

Move to the next location and repeat the procedure

03/25/25 36
Local Histogram Processing: Example

03/25/25 37
 Using Histogram Statistics for Image
Enhancement
Average Intensity L  1
M 1 N1
1
m  ri p (ri ) 
MN
  f ( x, y )
x 0 y 0
i 0
L 1
un (r )  (ri  m) n p (ri )
i 0

Variance L 1 M 1 N1
1
 u2 (r )  (ri     f ( x, y )  m 
2 2 2
m) p (ri ) 
i 0
MN x 0 y 0

03/25/25 38
Using Histogram Statistics for Image
Enhancement
Local average intensity
L 1
msxy  ri psxy (ri )
i 0

sxy denotes a neighborhood

Local variance
L 1
 2
s xy  (ri  msxy ) psxy (ri )
2

i 0

03/25/25 39
Using Histogram Statistics for Image
Enhancement: Example

 E  f ( x, y ), if msxy k0 mG and k1 G  sxy k 2 G

g ( x, y ) 
 f ( x, y ), otherwise

mG : global mean;  G : global standard deviation

k0 0.4; k1 0.02; k 2 0.4; E 4

03/25/25 40
 Spatial Filtering
linear spatial filtering using a 3X3 kernel
g(x, y), of the filter is the sum of products
of the kernel coefficients and the image pixels
encompassed by the kernel:
g(x,y)=w(-1,-1)f(x-1, y-1) + w(-1,0)f(x-1, y) +…
+ w(0,0)f(x, y) + …..+ w(1,1)f(x+1, y+1)

A spatial filter consists of (a)

a neighborhood, and (b) a
predefined operation

As coordinates x and y are varied, the

center of the kernel moves from pixel
to pixel, generating the filtered
03/25/25 41
image, g, in the process
 Spatial Filtering Contd…….

Linear spatial filtering of an image of size MxN with a

filter kernel of size m x n is given by the expression,
a b
g ( x, y )    w(s, t ) f ( x  s, y  t )
s  a t  b

m 2a  1; n 2b  1
x = 0, 1, 2, …..M-1, y = 0, 1, 2…N-1. a and b are
nonnegetive integers. Our focus is on kernels of odd size in
both coordinate directions. Smallest meaningful size 3X3.
where x and y are varied so that the center (origin) of the
kernel visits every pixel in f once

03/25/25 42
 Spatial Correlation & Convolution

Correlation consists of moving the center of a kernel over an image,

and computing the sum of products at each location.

Process of liner filtering is similar to a frequency domain

concept called Convolution.

Filter mask are sometimes called convolution masks or Kernel.

The mechanics of spatial convolution are the same, except that the
correlation kernel is rotated by 180°

Thus, when the values of a kernel are symmetric about its center,
correlation and convolution yield the same result.

a b
w( x, y ) f ( x, y )    w(s, t ) f ( x  s, y  t )
s  a t  b
03/25/25 43
 Spatial Convolution

The convolution of a filter w( x, y) of size m n

with an image f ( x, y ), denoted as w( x, y) f ( x, y)

a b
w( x, y ) f ( x, y )    w( s, t ) f ( x  s, y  t )
s  a t  b

03/25/25 44
Spatial Correlation & Convolution
Contd…

Response R of an Correlation consists of moving the center of a

kernel over an image, and computing the sum of products at each
location.
Response of the filter:
R = w1z1 + w2z2 + …………………….. wmnzmn

W is the mask coefficients, mn is total no. of coefficients, z’s are the

values of image grey levels. For 3X3 general mask shown below, the
response at any point (x, y) isW1 W2 W3 z1 z2 z3

W4 W5 W6 z4 z5 z6

W7 W8 W9 z7 z8 z9

03/25/25 45
03/25/25 46
 Construction of Spatial Filter
Kernels
First approach is based on formulating filters based on
mathematical properties. For example, a filter that computes the
average of pixels in a neighborhood blurs an image. Computing an
average is analogous to integration. Conversely, a filter that
computes the local derivative of an image sharpens the image.

A second approach is based on sampling a 2-D spatial function

whose shape has a desired property. Gaussian function can be
used to construct a weighted-average (lowpass) filter. These 2-D
spatial functions sometimes are generated as the inverse Fourier
transform of 2-D filters specified in the frequency domain.

A third approach is to design a spatial filter with a specified

frequency response. This approach falls in the area of digital filter
design.

(Page 238)
03/25/25 47
 Separable Filter Kernels

A spatial filter kernel is a matrix, and a separable kernel is a matrix

that can be expressed as the outer product of two vectors. For
example, the 2 X 3 kernel is separable because it can be expressed
as the outer product of the vectors

A separable kernel of size m x n can be expressed as the outer

product of two vectors
(Page 238)
03/25/25 48
 Separable Filter Kernels

The importance of separable kernels lies in the computational

advantages that result from the associative property of
convolution. If we have a kernel w , that can be decomposed into
two simpler kernels, such that
w = w1*w2 then it follows from the commutative and associative
properties

This equation says that convolving a separable kernel with an

image is the same as convolving w1 with f first, and then
convolving the result with w2.
03/25/25 49
 Smoothing Spatial Filters

Smoothing (Averaging) filters are used for blurring and

for noise reduction.
Blurring is used in removal of small details and
bridging of small gaps in lines or curves. pixel
regions that are small with respect to the size of
the filter kernel
Noise Reduction: accomplished with the help of
blurring with linear/nonlinear filters.
used to reduce sharp transitions in intensity. Because
random noise typically consists of sharp transitions in
intensity, an obvious application of smoothing is noise
reduction. Eg. Reduction of Aliasing effect.
Smoothing filters are used in combination with other
techniques for image enhancement.
03/25/25 50
(Page 238)
 Smoothing Spatial Filters Contd..
Smoothing spatial filters include linear filters and nonlinear
filters.

Linear spatial filtering consists of convolving an image with a

filter kernel. Convolving a smoothing kernel with an image
blurs the image, with the degree of blurring being determined
by the size of the kernel and the values of its coefficients.

Lowpass filters are fundamental, in the sense that other

important filters, including sharpening (highpass), bandpass,
and bandreject filters, can be derived from lowpass filters.

We discuss in this section lowpass filters based on Box and

Gaussian kernels, both of which are separable
(Page 239)
03/25/25 51
 Two Smoothing Averaging Filter
Masks

Standard Filter Mask Weighted average Filter Mask

The simplest, separable lowpass filter

kernel is the box kernel, whose
coefficients have the same value
(typically normalizing
1). constant in front, whose
value is 1 divided by the sum of the
03/25/25 52
values of the
coefficients
Two Smoothing Averaging Filter
Masks Contd.

Weighted average Filter Mask

Standard Filter Mask

average value of an area of constant intensity

would equal that intensity in the filtered image,
as it should
normalizing the kernel in this way prevents introducing a
bias during filtering; that is, the sum of the pixels in the
03/25/25 53
original and filtered images will be the same. Rank of kernels
is 1, so separable.
 Spatial Smoothing Linear Filters

The general implementation for filtering an M N image

with a weighted averaging filter of size m n is given
a b

  w(s, t ) f ( x  s, y  t )
g ( x, y )  s  a t  b a b

  w(s, t )
s  a t  b

where m 2a  1, n 2b  1.

03/25/25 54
03/25/25 55
Example: Gross Representation of Objects

03/25/25 56
 Order-statistic (Nonlinear) Filters

— Nonlinear

— Response is based on ordering (ranking) the

pixels contained in the filter mask

— Replacing the value of the center pixel with the

value determined by the ranking result

E.g., median filter, max filter, min filter

03/25/25 57
 Median Filter

Order statistic filter / median filter Objective: Replace the

valve of the pixel by the median of the gray levels or intensity
values in the neighborhood of that pixel. (the original value of
the pixel is included in the computation of the median).

It has excellent noise reduction capability.

Produces less blurring compare to linear smoothing filters
Effective for Impulse noise/salt and pepper noise.

Although the median filter is by far the most useful order-

statistics filter in image processing, it is by no means the only
one.
03/25/25 58
Example: Use of Median Filtering for Noise
Reduction

03/25/25 59
 Median Filter Contd.

03/25/25 60
 Max Filter

The median represents the 50th percentile of a ranked set of

numbers. Using the 100th percentile results in the so-called

Max filter, which is useful in finding the brightest points

in an image.
The response of a 3*3 max filter is given by
R = max [ zk| k=1, 2, ,… 9]

03/25/25 61
 Min Filter

The 0th percentile filter is the Min filter, used for the
opposite purpose.
Example nonlinear spatial filters
–Median filter: Computes the median gray-level value of the
neighborhood. Used for noise reduction.
– Max filter: Used to find the brightest points in an image
–Min filter: Used to find the dimmest points in an image
R = max{z | k =1,2,...,9}
R = min{z | k =1,2,...,9}
03/25/25 62
 Sharpening Spatial Filters
(High Pass Filtering)

Image blurring could be accomplished in the spatial domain

by pixel averaging (smoothing) in a neighborhood.

Sharpening: to enhance the details which has been blurred.

Highlight the finite details.

Image blurring means pixel averaging. Since averaging is

analogous to integration, it is logical to conclude that
sharpening could be accomplished by spatial differentiation.

03/25/25 63
 Sharpening Spatial Filters Contd.
(High Pass Filtering)

This section deals with various ways of defining and

implementing operators for Image sharpening by digital
differentiation.

Fundamentally, the strength the response of a derivative

operator is proportional to the degree of discontinuity of
the image at the point at which the operator is applied.
Advantage:
Thus, image differentiation enhances edges and other
discontinuities (such as noise) and deemphasizes areas with
slowly varying gray- level values.
03/25/25 64
 Sharpening Spatial Filters Contd.

► Foundation

► Laplacian Operator

► Unsharp Masking and Highboost Filtering

► Using First-Order Derivatives for Nonlinear Image

Sharpening

03/25/25 65
 Sharpening Spatial Filters: Foundation

► The first-order derivative of a one-dimensional

function f(x) is the difference

f
 f ( x  1)  f ( x)
x
► Derivatives of digital fn: Defined in terms of
differences
► Properties:
First derivative
(1)must be zero in flat segments (areas of constant gray-level
values);

(2)must be nonzero at the onset of a gray-level Step or Ramp; and

03/25/25 66
(3)must be nonzero along Ramps.
 Sharpening Spatial Filters: Foundation
Contd.

► The second-order derivative of f(x) as the difference

2 f
2
 f ( x  1)  f ( x  1)  2 f ( x)
x
Similarly, any definition of a second derivative
 must be zero in flat areas;

 must be nonzero at the onset and end of a gray-level

Step or Ramp;

 must be zero along Ramps of constant slope

Dealing with digital quantities: Max gray level is finite and

adjacent one also finite. Shortest distance over which
03/25/25
change can occur is b/w adjacent pixels. 67
First derivative
(1)must be zero in flat segments (areas of Second derivative
constant gray-level values); (1) must be zero in flat areas;
(2)must be nonzero at the onset of a gray-level (2) must be nonzero at the onset and
step or ramp; and end of a gray-level step or ramp;
(3)must be nonzero along ramps (3) must be zero along ramps of
03/25/25 constant slope. 68
03/25/25 69
 Sharpening Spatial Filters: Laplace
Operator

The second-order isotropic derivative operator is the

Laplacian for a function (image) f(x,y)
2 2
2  f  f 2 f
 f  2  2 x 2
 f ( x  1)  f ( x  1)  2 f ( x)
x y
2 f
► In x direction, 2  f ( x  1, y )  f ( x  1, y )  2 f ( x, y )
x
2 f
► In y direction, 2  f ( x, y  1)  f ( x, y  1)  2 f ( x, y )
y
2 f  f ( x  1, y )  f ( x  1, y )  f ( x, y  1)  f ( x, y  1)
- 4 f ( x, y )
03/25/25 70
 Sharpening Spatial Filters: Laplace
Operator Contd.

03/25/25 71
 Sharpening Spatial Filters: Laplace
Operator Contd.

Image sharpening in the way of using the Laplacian:

g ( x, y )  f ( x, y )  c  2 f ( x, y ) 
where,
f ( x, y ) is input image,
g ( x, y ) is sharpenend images,
c -1 if 2 f ( x, y ) corresponding to Fig. 3.37(a) or (b)
and c 1 if either of the other two filters is used.

03/25/25 72
LAPLACIA + ADDITION WITH ORIGINAL IMAGE
N DIRECTLY

73
03/25/25 74
Unsharp Masking and Highboost Filtering
► Unsharp masking
Sharpen images consists of subtracting an unsharp
(smoothed) version of an image from the original image
e.g., printing and publishing industry

► Steps
1. Blur the original image

2. Subtract the blurred image from the original

3. Add the mask to the original

03/25/25 75
Unsharp Masking and Highboost Filtering
Let f ( x, y ) denote the blurred image, unsharp masking is
g mask ( x, y )  f ( x, y )  f ( x, y )
Then add a weighted portion of the mask back to the original
g ( x, y )  f ( x, y )  k * g mask ( x, y ) k 0

when k  1, the process is referred to as highboost filtering.

03/25/25 76
Unsharp Masking: Demo

03/25/25 77
Unsharp Masking and Highboost Filtering:
Example

03/25/25 78
Image Sharpening based on First-Order
Derivatives
For function f ( x, y ), the gradient of f at coordinates ( x, y )
is defined as
 f 
 g x   x 
f grad( f )    
 g y   f 
 y 

The magnitude of vector f , denoted as M ( x, y )

Gradient Image M ( x, y ) mag(f )  g x 2  g y 2
03/25/25 79
Image Sharpening based on First-Order
Derivatives
The magnitude of vector f , denoted as M ( x, y )
M ( x, y ) mag(f )  g x 2  g y 2

M ( x, y ) | g x |  | g y |

z1 z2 z3
M ( x, y ) | z8  z5 |  | z6  z5 |
z4 z5 z6
z7 z8 z9
03/25/25 80
Image Sharpening based on First-Order
Derivatives

Roberts Cross-gradient Operators

M ( x, y ) | z9  z5 |  | z8  z6 |

Sobel Operators
M ( x, y ) | ( z7  2 z8  z9 )  ( z1  2 z2  z3 ) |
z1 z2 z3  | ( z3  2 z6  z9 )  ( z1  2 z4  z7 ) |
z4 z5 z6
z7 z8 z9
03/25/25 81
Image Sharpening based on First-Order
Derivatives

03/25/25 82
Example

03/25/25 83
Example:

Combining
Spatial
Enhancement
Methods

Goal:

Enhance the
image by
sharpening it
and by bringing
out more of the
skeletal detail
03/25/25 84
Example:

Combining
Spatial
Enhancement
Methods

Goal:

Enhance the
image by
sharpening it
and by bringing
out more of the
skeletal detail
03/25/25 85