Digital Image Processing

Intensity Transformation and

Spatial Filtering
 Spatial Domain vs. Transform Domain

► Spatial domain
image plane itself, directly process the intensity values of the image plane

► Transform domain
process the transform coefficients, not directly process the intensity values of
the image plane

11/14/22 2
 Spatial Domain Process

g ( x, y )  T [ f ( x, y )])
f ( x, y ) : input image
g ( x, y ) : output image
T : an operator on f defined over
a neighborhood of point ( x, y )

11/14/22 3
 Spatial Domain Process

11/14/22 4
 Spatial Domain Process

Intensity transformation function

s  T (r )

11/14/22 5
 Some Basic Intensity Transformation Functions

11/14/22 6
 Image Negatives

Image negatives
s  L 1 r

11/14/22 7
 Example: Image Negatives

Small
lesion

11/14/22 8
 Log Transformations

Log Transformations
s  c log(1  r )

11/14/22 9
 Example: Log Transformations

11/14/22 10
 Power-Law (Gamma) Transformations


s  cr

11/14/22 11
 Example: Gamma Transformations

11/14/22 12
 Example: Gamma Transformations
Cathode ray tube
(CRT) devices have an
intensity-to-voltage
response that is a
power function, with
exponents varying
from approximately
1.8 to 2.5

11/14/22 13
 Example: Gamma Transformations

11/14/22 14
 Piecewise-Linear Transformations

► Contrast Stretching
— Expands the range of intensity levels in an image so that it spans the full intensity range of
the recording medium or display device.

► Intensity-level Slicing
— Highlighting a specific range of intensities in an image often is of interest.

11/14/22 15
11/14/22 16
 Highlight the major
blood vessels and
study the shape of the
flow of the contrast
medium (to detect
blockages, etc.)

Measuring the actual

flow of the contrast
medium as a function
of time in a series of
11/14/22
images 17
 Bit-plane Slicing

11/14/22 18
 Bit-plane Slicing

11/14/22 19
 Bit-plane Slicing

11/14/22 20
► In terms of intensity transformation functions, it is not difficult to show that the binary image for the 8th bit plane of
an 8-bit image can be obtained by processing the input image with a thresholding intensity transformation function
that maps all intensities between 0 and 127 to 0 and maps all levels between 128 and 255 to 1.
 It is left as an exercise (Problem 3.4) to obtain the intensity transformation functions for generating the other bit planes.

► Decomposing an image into its bit planes is useful for analyzing the relative importance of each bit in the image, a
process that aids in determining the adequacy of the number of bits used to quantize the image. Also, this type of
decomposition is useful for image compression (the topic of Chapter 8), in which fewer than all planes are used in
reconstructing an image.

► The reconstruction is done by multiplying the pixels of the nth plane by the constant pow(2, n-1)
► This is nothing more than converting the nth significant binary bit to decimal.
► Each plane used is multiplied by the corresponding constant, and all planes used are added to obtain the gray scale
image. Thus, to obtain Fig. 3.15(a), we multiplied bit plane 8 by 128, bit plane 7 by 64, and added the two planes.
► Although the main features of the original image were restored, the reconstructed image appears flat, especially in the
background. This is not surprising because two planes can produce only four distinct intensity levels.

11/14/22 21
 Histogram Processing

► Histogram Equalization

► Histogram Matching

► Local Histogram Processing

► Using Histogram Statistics for Image Enhancement

11/14/22 22
 Histogram Processing

Histogram h( rk )  nk
rk is the k th intensity value
nk is the number of pixels in the image with intensity rk

nk
Normalized histogram p( rk ) 
MN
nk : the number of pixels in the image of
size M  N with intensity rk

11/14/22 23
11/14/22 24
 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.

11/14/22 25
 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.

11/14/22 26
 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.

11/14/22 27
 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 
11/14/22 28
 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.

11/14/22 29
 Example
r
s  T (r )  ( L  1)  pr ( w)dw
0

r 2w
 ( L  1)  2
dw
0 ( L  1)
2
r

L 1

11/14/22 30
 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
11/14/22 31
 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.

11/14/22 32
 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
11/14/22 33
 Example: Histogram Equalization

11/14/22 34
11/14/22 35
11/14/22
 Question
Is histogram equalization always good?

No

11/14/22 37
 Histogram Matching
Histogram matching (histogram specification)
— generate a processed image that 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

Define a random variable z with the probability

z
G ( z )  ( L  1)  pz (t ) dt  s
0
11/14/22 38
 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)
 “histogram equalization transformation for the specified histogram”
z
G ( z )  ( L  1)  pz (t )dt  s
0

► Mapping from s to z
z  G 1 ( s )

11/14/22 39
 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
 3
, for 0  z  ( L -1)
pz ( z )   ( L  1)
 0, otherwise

11/14/22 40
 Histogram Matching: Example

Find the histogram equalization transformation for the input image

2
r r 2w r
s  T (r )  ( L  1)  pr ( w)dw  ( L  1)  2
dw 
0 0 ( L  1) L 1

Find the histogram equalization transformation for the specified histogram

z z 3t 2 z3
G ( z )  ( L  1)  pz (t )dt  ( L  1)  3
dt  2
s
0 0 ( L  1) ( L  1)
The transformation function
2 1/3
2 1/3  2 r  2 1/3
z  ( L  1) s   ( L  1)   ( L  1) r 
 L  1 
11/14/22 41
 2 1/3
2 1/3  2 r  2 1/3
z   ( L  1) s   ( L  1)   ( L  1) r 
 L  1 

► Thus, squaring the value of each pixel in

the original image, multiplying the result
by (L-1) and raising the product to the
power 1/3 will yield an image whose
intensity levels, z, have the specified PDF.

11/14/22 42
 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
zq  G 1 ( sk )
11/14/22 43
 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.

11/14/22 44
 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

11/14/22 45
 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 s 5 s 6 s7

11/14/22 46
 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
11/14/22
6 47

7
 Example: Histogram Matching
rk  zq
03
1 4
25
36
47
57
67
77
11/14/22 48
 Example: Histogram Matching

11/14/22 49
 Example: Histogram Matching

11/14/22 50
 Example: Histogram Matching

11/14/22 51
 Example: Histogram Matching

11/14/22 52
 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

11/14/22 53
 Local Histogram Processing: Example

11/14/22 54
 Using Histogram Statistics for Image Enhancement

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

11/14/22 55
 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

11/14/22 56
 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

11/14/22 57
 Spatial Filtering

A spatial filter consists of (a) a neighborhood, and (b) a

predefined operation

Linear (filter that implements a linear operation) spatial

filtering of an image of size MxN with a filter of size mxn is
given by the expression
a b
g ( x, y )    w(s, t ) f ( x  s, y  t )
s  a t  b

11/14/22 58
 Spatial Filtering

11/14/22 59
 Spatial Correlation

The correlation 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

Sum of the product

https://setosa.io/ev/image-kernels/

11/14/22 60
 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

Same thing but convolution is associative !!

11/14/22 61
11/14/22 62
11/14/22 63
11/14/22 64
11/14/22 65
 Smoothing Spatial Filters

Smoothing 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

Smoothing spatial filters include linear filters and nonlinear

filters.

11/14/22 66
 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.

11/14/22 67
 Two Smoothing Averaging Filter Masks

11/14/22 68
11/14/22 69
 Example: Gross Representation of Objects

11/14/22 70
 Order-statistic (Nonlinear) Filters

— Nonlinear

— 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

11/14/22 71
 Example: Use of Median Filtering for Noise Reduction

11/14/22 72
 Sharpening Spatial Filters

► Foundation

► Laplacian Operator

► Unsharp Masking and Highboost Filtering

► Using First-Order Derivatives for Nonlinear Image

Sharpening — The Gradient

11/14/22 73
 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

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

2 f
2
 f ( x  1)  f ( x  1)  2 f ( x)
x

11/14/22 74
11/14/22 75
 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
 f  2  2
x y
2 f
2
 f ( x  1, y )  f ( x  1, y )  2 f ( x, y )
x
2 f
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 )
11/14/22 76
 Sharpening Spatial Filters: Laplace Operator

11/14/22 77
 Sharpening Spatial Filters: Laplace Operator

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.

11/14/22 78
11/14/22 79
 Image scaling (Ch2)

(OR) y= ((x-min)/(max-min))*255

11/14/22 80
 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 (creates a mask)

3. Add the mask to the original

11/14/22 81
 Unsharp Masking: Demo

11/14/22 82
 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.

11/14/22 83
 Unsharp Masking and Highboost Filtering: Example

11/14/22 84
 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
11/14/22 85
 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
11/14/22 86
 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 ) |
 | ( z3  2 z6  z9 )  ( z1  2 z4  z7 ) |

z1 z2 z3 The idea behind using a weight value of 2

in the center coefficient is to achieve some
z4 z5 z6 smoothing by giving more importance to the
center point
z7 z8 z9
11/14/22 87
 Image Sharpening based on First-Order Derivatives
The coefficients in all the
masks shown in Fig.3.41
sum to 0 . . WHY?

Reason: they would give

a response of 0 in an
area of constant
intensity, as is expected
of a derivative operator.

11/14/22 88
Example

89

11/14/22
 Example:

Combining
Spatial
Enhancement
Methods

Goal:

Enhance the
image by
sharpening it
and by bringing
out more of the
skeletal detail

11/14/22 90
 Example:

Combining
Spatial
Enhancement
Methods

Goal:

Enhance the
image by
sharpening it
and by bringing
out more of the
skeletal detail

11/14/22 91