Digital Image Processing

Image Enhancement in the

Spatial Domain

Dr. A. NAGARAJA RAO

Objective
The principal objective of enhancement is to process an
images so that the result is more suitable than the original
image for a SPECIFIC application

Category of image enhancement

• Spatial domain
• Frequency domain

Dr. A. NAGARAJA RAO

Background
The spatial domain process can be denoted as
g(x,y) = T [ f(x,y) ]

Dr. A. NAGARAJA RAO

Point Processing,
Gray-Level Transformation Function

s = T (r)

Contrast Thresholding
stretching

Dr. A. NAGARAJA RAO

Basic Gray Level Transformation

Dr. A. NAGARAJA RAO

Image Negative

s=L-1-r

Dr. A. NAGARAJA RAO

Log Transformations
s = c log(1+r)

Pixel values dynamic range=[0 - 1.5×106]

Dr. A. NAGARAJA RAO
 Power-Law Transformations

s = cr

s = r2.5
r = [1 10 20 30 40 210 220 230 240 250 255]
s( = 2.5) = [0 0 0 1 2 157 176 197 219 243 255]
s( =.4) = [28 70 92 108 122 236 240 245 249 253 255]
Dr. A. NAGARAJA RAO
Power-Law Transformations - Gamma Correction

Dr. A. NAGARAJA RAO

Power-Law Transformations
c=1
 = 0.6

c=1
 = 0.4 c=1
 = 0.3
Dr. A. NAGARAJA RAO
Power-Law Transformations

c=1
=3

c=1 c=1
=4 =5

Dr. A. NAGARAJA RAO

Piecewise-Linear Transformation
Functions

• Advantage
• Arbitrarily complex
• Disadvantage
• More user input
• Type of Transformations
• Contrast stretching
• Gray-level slicing
• Bit-plane slicing

Dr. A. NAGARAJA RAO

 Contrast Stretching
Objective: Increase the dynamic range of the gray levels in the
image

Causes for poor image

-Poor illumination
-Wrong lens aperture
-Wrong shutter speed
-Lack of dynamic range
in the imaging sensor

Dr. A. NAGARAJA RAO

If r1=s1 and r2=s2 then T is linear function
If r1=r2 , s1=0 and s2=L-1 then T is thresholding function
• generally the condition we assume r1<= r2 and s1<=s2
then the function is single valued and monotonically
increasing

Contrast stretching:
(r1,s1)=(rmin,0) , (r2,s2)=(rmax,L-1)
Thresholding:
(r1,s1)=(k,0) , (r2,s2)=(k,L-1)

Dr. A. NAGARAJA RAO

 Image (s ( after applying T (contrast
Image (r( stretching)

Notice that the intensity transformation function T,

made the pixels with dark intensities darker and the
bright ones even more brighter, this is called
contrast stretching>

Dr. A. NAGARAJA RAO

Contrast Stretching
Example 1
For image with intensity range [50 - 150]
What should (r1,s1) and (r2,s2) be to increase the dynamic
range of the image to [0 - 255]?

Dr. A. NAGARAJA RAO

r
 piecewise Linear transformation functions.
Exercise on Contrast stretching and thresholding

Exercise:

the following matrix represents the pixels values of a 8-bit image (r) , apply
thresholding transform assuming that the threshold m=95, find the resulting image
pixel values.
Image (r)

solution: 110 120 90 130

Image (s) 91 94 98 200

90 91 99 100

82 96 85 90

Dr. A. NAGARAJA RAO

Gray-Level Slicing
Objective: Highlighting a specific range of gray levels in an image.

Dr. A. NAGARAJA RAO

Dr. A. NAGARAJA RAO
Bit-Plane Slicing
255 138 30
65 12 201
180 111 85
LSB MSB
MSB plane 255 1 1 1 1 1 1 1 1
1 1 0 138 0 1 0 1 0 0 0 1
0 0 1 30 1 1 1 1 1 0 0 0
1 0 0 65 1 0 0 0 0 0 1 0
12 0 0 1 1 0 0 0 0
LSB plane 201 1 0 0 1 0 0 1 1
1 0 1 183 1 1 1 0 1 1 0 1
1 0 1 111 1 1 1 1 0 1 1 0
1 1 1 1 RAO0
85 Dr. A. NAGARAJA 1 0 1 0 1 0
Dr. A. NAGARAJA RAO
Useful for compression.

Reconstruction is obtained by:

N
I (i, j )   2n 1 I n (i, j )
n 1

Dr. A. NAGARAJA RAO

Dr. A. NAGARAJA RAO
Bit-Plane Slicing

Dr. A. NAGARAJA RAO

 Bit Plane Slicing (cont…)

[10000000] [01000000]

[00100000] [00001000]

[00000100] [00000001]

Dr. A. NAGARAJA RAO

Histogram Processing

Histogram
The histogram of a digital image with gray levels in
the range [0, L-1] is a discrete function
h(rk) = nk,
where rk is the kth gray level and nk is the number of
pixels in the image having gray level rk.

Normalized Histogram
Dividing each value of the histogram by the total number
of pixels in the image, denoted by n.
p(rk) = nk /n.
Normalized histogram provide useful image statistics.
Dr. A. NAGARAJA RAO
 Image Histograms
The histogram of an image shows us the
distribution of grey levels in the image
Massively useful in image processing, especially in
segmentation
Frequencies

Grey Levels
Dr. A. NAGARAJA RAO
Histogram Examples

Dr. A. NAGARAJA RAO

Histogram Examples (cont…)

Dr. A. NAGARAJA RAO

Histogram Processing

Dr. A. NAGARAJA RAO

Histogram Examples (cont…)

Dr. A. NAGARAJA RAO

Histogram Examples (cont…)

Dr. A. NAGARAJA RAO

Histogram Examples (cont…)

Dr. A. NAGARAJA RAO

Histogram Examples (cont…)

Dr. A. NAGARAJA RAO

Histogram Examples (cont…)

Dr. A. NAGARAJA RAO

Histogram Examples (cont…)

Dr. A. NAGARAJA RAO

Histogram Examples (cont…)

Dr. A. NAGARAJA RAO

Histogram Examples (cont…)

Dr. A. NAGARAJA RAO

 Histogram Examples (cont…)
• A selection of images and
their histograms
• Notice the relationships
between the images and
their histograms
• Note that the high contrast
image has the most
evenly spaced histogram

Dr. A. NAGARAJA RAO

 Histogram Equalisation
• Spreading out the frequencies in an image (or
equalising the image) is a simple way to
improve dark or washed out images.
• At first, the continuous case will be studied:
– r is the intensity of the image in [0, L-1].
– we focus on transformations s=T(r):
• T(r) is strictly monotonically increasing.
• T(r) must satisfy:
0  T (r )  L  1, for 0  r  L  1
 Histogram Equalisation (cont...)
• The condition for T(r) to be monotonically
increasing guarantees that ordering of the output
intensity values will follow the ordering of the
input intensity values (avoids reversal of
intensities).
• If T(r) is strictly monotonically increasing then
the mapping from s back to r will be 1-1(one-to-
one).
• The second condition (T(r) in [0,1]) guarantees
that the range of the output will be the same as
the range of the input.
 Histogram Equalisation (cont...)
• We can view intensities r and s as random
variables and their histograms as probability
density functions (pdf) pr(r) and ps(s).
• Fundamental result from probability theory:
– If pr(r) and T(r) are known and T(r) is continuous
and differentiable, then

1 dr
ps ( s)  pr (r )  pr (r )
ds ds
dr
 Histogram Equalisation (cont...)
• The pdf of the output is determined by the pdf
of the input and the transformation.
• This means that we can determine the
histogram of the output image.
• A transformation of particular importance in
image processing is the cumulative distribution
function (CDF) of a random variable:
r
s  T (r )  ( L  1)  pr ( w) dw
0
 Histogram Equalisation (cont...)
• It satisfies the first condition as the area under
the curve increases as r increases.
• It satisfies the second condition as for r=L-1 we
have s=L-1.
• To find ps(s) we have to compute

r
ds dT (r ) d
  ( L  1)  pr ( w) dw  ( L  1) pr (r )
dr dr dr 0
 Histogram Equalisation (cont...)
Substituting this result:
ds
 ( L  1) pr (r )
dr
to
dr
ps ( s)  pr (r ) Uniform pdf
ds
yields
1 1
ps ( s)  pr (r )  , 0  s  L 1
( L  1) pr (r ) L 1
 Histogram Equalisation (cont...)
The formula for histogram equalisation in the
discrete case is given
k
( L  1) k
sk  T (rk )  ( L  1) pr (rj )   nj
j 0 MN j 0
where
• rk: input intensity
• sk: processed intensity
• nj: the frequency of intensity j
• MN: the number of image pixels.
 Histogram Equalisation (cont...)
Example
A 3-bit 64x64 image has the following intensities:
k
sk  T (rk )  ( L  1) pr (rj )
j 0

Applying histogram equalization:

0
s0  T (r0 )  7 pr (rj )  7 pr (r0 )  1.33
j 0
1
s1  T (r1 )  7 pr (rj )  7 pr (r0 )  7 pr (r1 )  3.08
j 0
 Histogram Equalisation (cont...)
Example
Rounding to the nearest integer:

s0  1.33  1 s1  3.08  3 s2  4.55  5 s3  5.67  6

s4  6.23  6 s5  6.65  7 s6  6.86  7 s7  7.00  7
 Histogram Equalization (cont…)
Example
Notice that due to discretization, the resulting
histogram will rarely be perfectly flat. However, it will
be extended.

Dr. A. NAGARAJA RAO

Equalisation Transformation Function

Dr. A. NAGARAJA RAO

Equalisation Examples
1

Dr. A. NAGARAJA RAO

 Equalisation Transformation Functions

The functions used to equalise the images in the

previous example

Dr. A. NAGARAJA RAO

Equalisation Examples
2

Dr. A. NAGARAJA RAO

 Equalisation Transformation Functions

The functions used to equalise the images in the

previous example

Dr. A. NAGARAJA RAO

Equalisation Examples (cont…)
3

Dr. A. NAGARAJA RAO

 Equalisation Transformation Functions

The functions used to equalise the images in the

previous examples

Dr. A. NAGARAJA RAO

Local Histogram Enhancement

Dr. A. NAGARAJA RAO

Histogram Statistics for Image Enhanc.
Contrast manipulation using local statistics, such as the mean and
variance, is useful for images where part of the image is acceptable,
but other parts may contain hidden features of interest.

Dr. A. NAGARAJA RAO

Histogram Statistics for Image Enhanc.
Let (x,y) be the coordinates of a pixel in an image, and let Sxy
denote a neighborhood (sub-image) of specified size, centered at
(x,y).

mS xy   rs,t p(rs,t )
( s ,t )S xy

 S xy
2
  [rs,t  mS xy
]2
p ( rs ,t ).
( s ,t )S xy

The local mean and variance are the decision factors to whether apply
local enhancement or not.

Dr. A. NAGARAJA RAO

 Histogram Statistics for Image Enhanc.
MG: Global mean
DG: Global standard
deviation
E, k0, k1, k2: Specified
parameters

E. f ( x, y) if mS xy  k0 M G AND k1DG   S xy  k2 DG

g ( x, y)  
 f ( x, y) otherwise E = 4,
k0 = 0.4,
Size of local area = 3  3 k1= 0.02,
Dr. A. NAGARAJA RAO
k2 = 0.4
 Histogram Statistics for Image Enhanc.

(x,y) E

(x,y)

Image Mean
125 135 145 135 .. 174 164
168 175 158 149 .. 187
210 231 215 129 ..
187 192 145 200 .. Dr. A. NAGARAJA RAO
Histogram Statistics for Image Enhanc.

Enhanced Image Original Image

Dr. A. NAGARAJA RAO

Enhancement Using Arithmetic/Logic Op.
Arithmetic/logic operations involving images are performed on a
pixel-by-pixel basis between two or more images.

Arithmetic Operations
Addition, Subtraction, Multiplication, and Division

Logic Operations
AND, OR, NOT

Dr. A. NAGARAJA RAO

Enhancement Using AND and OR Logic Op.

AND

OR

Logic operations are performed on the binary representation of the

pixel intensities
Dr. A. NAGARAJA RAO
 Enhancement Using Arithmetic Op. _ SUB

Original
4
Upper-order
Bit

HE of
Error Error

Dr. A. NAGARAJA RAO

Enhancement Using Arithmetic Op. _ SUB
Mask Mode Radiography

Problem:
The pixel intensities in the difference image can range from -255 to 255.
Solutions:
1) Add 255 to every pixel and then divide by 2.
2) Add the minimum value of the pixel intensity in the difference image
to every pixel and then divide by 255/Max. Max is the maximum
pixel value in the modified difference image.
Dr. A. NAGARAJA RAO
Enhancement Using Arithmetic Op. Averaging
Original image Noise with zero mean

g ( x , y )  f ( x, y )   ( x, y )
1 k
g ( x , y )   g i ( x, y )
k i 1
E g ( x, y )  f ( x, y )
1 2
 g ( x, y )    ( x, y )
2

k
Dr. A. NAGARAJA RAO
 Enhancement Using Arithmetic Op. Averaging

Gaussian Noise
f(x,y) g(x,y) mean = 0
variance = 64

K=8 K = 16

x
-  +
K = 128
K = 32
Dr. A. NAGARAJA RAO
 Enhancement Using Arithmetic Op. Averaging

K=8

Difference images
between original K = 16
image and images
obtained from Notice the
averaging. mean and
K = 32 variance
of the
difference
images
K = 128 decrease as K
Dr. A. NAGARAJA RAO increases.
Basics of Spatial Filtering - Linear

Spatial filtering are filtering

operations performed on the
pixel intensities of an image
and not on the frequency
components of the image.

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

Dr. A. NAGARAJA RAO

Basics of Spatial Filtering
Response, R, of an m  n mask at any point (x, y)

mn
R   wi zi
i 1

Special consideration is given when the center of the filter

approach the boarder of the image.

Dr. A. NAGARAJA RAO

Nonlinear of Spatial Filtering

Nonlinear spatial filters operate on neighborhoods, and the mechanics

of sliding a mask past an image are the same as was just outlined. In
general however, the filtering operation is based conditionally on the
values of the pixel in the neighborhood under consideration, and they
do not explicitly use coefficients in the sum-of products manner
described previously.

Example
Computation for the median is a nonlinear operation.

Dr. A. NAGARAJA RAO

Smoothing Spatial Filtering - Linear
Averaging (low-pass) Filters
Smoothing filters are used
- Noise reduction
- Smoothing of false contours
- Reduction of irrelevant detail

Undesirable side effect of smoothing filters

- Blur edges

Weighted average filter

reduces blurring in the
smoothing process.

Box Weighted
filter average
Dr. A. NAGARAJA RAO
 Smoothing Spatial Filtering _ Linear
Averaging (low-pass) Filters
n = filter size

n=3

n=5 n=9

n = 35

n = 15
Dr. A. NAGARAJA RAO
Smoothing Spatial Filtering
Averaging & Threshold
filter size Thrsh = 25% of
n = 15 highest intensity

Dr. A. NAGARAJA RAO

Smoothing Spatial Filtering
Order Statistic Filters
Order-statistics filters are nonlinear spatial filters whose
response is based on ordering (ranking) the pixels contained
in the image area encompassed by the filter, and then
replacing the value of the center pixel with the value
determined by the ranking result.

3  3 Median filter [10 125 125 135 141 141 144 230 240] = 141
3  3 Max filter [10 125 125 135 141 141 144 230 240] = 240
3  3 Min filter [10 125 125 135 141 141 144 230 240] = 10

Median filter eliminates isolated clusters of pixels that are light or

dark with respect to their neighbors, and whose area is less than
n2/2.
Dr. A. NAGARAJA RAO
Order Statistic Filters
n=3 n=3
Average Median
filter filter

Dr. A. NAGARAJA RAO

 Sharpening Spatial Filters
The principal objective of sharpening is to highlight fine detail
in an image or to enhance detail that has been blurred.

1 9

9 1
zi
Image Blurred Image
1 9

9 1
z

– Sharpening could be accomplished by spatial differentiation

– Image differentiation enhances edges and other discontinuities
(such as noise) and deemphasizes areas with slowly varying gray-
level values.
The derivatives of a digital function are defined in terms of differences.

Dr. A. NAGARAJA RAO

Sharpening Spatial Filters
Requirements for digital derivative
First derivative
1) Must be zero in flat segment
2) Must be nonzero along ramps.
3) Must be nonzero at the onset of a gray-level step or ramp
Second derivative
1) Must be zero in flat segment
2) Must be zero along ramps.
3) Must be nonzero at the onset and end of a gray-level step or
ramp
f
 f ( x  1)  f ( x)
x
2 f
 f ( x  1)  f ( x  1)  2 f ( x)
x Dr. A. NAGARAJA RAO
2
Sharpening Spatial Filters

Dr. A. NAGARAJA RAO

Sharpening Spatial Filters
Comparing the response between first- and second-ordered
derivatives:
1) First-order derivative produce thicker edge
2) Second-order derivative have a stronger response to fine
detail, such as thin lines and isolated points.
3) First-order derivatives generally have a stronger response
to a gray-level step
4) Second-order derivatives produce a double response at
step changes in gray level.

In general the second derivative is better than the first

derivative for image enhancement. The principle use of first
derivative is for edge extraction.

Dr. A. NAGARAJA RAO

Use of First Derivative for Edge Extraction
Gradient
First derivatives in image processing are implemented
using the magnitude of the gradient.
t
 f f 
f   z1 z2 z3
 x y 
z4 z5 z6
2 0.5
 f   f  
2
z7 z8 z9
f  mag (f )         Gx  G y
 x   y  
Roberts operator
Gx = (z9-z5) and Gy = (z8 - z6)
Sobel operator
Gx = (z3+2z6 +z9) - (z1+2z4+z7) and
Gy = (z7+2z8+z9) - (z1+2z2+z3)
Dr. A. NAGARAJA RAO
Use of First Derivative for Edge Extraction
Gradient

Robert operator

Sobel operators
Dr. A. NAGARAJA RAO
Use of First Derivative for Edge Extraction
Gradient

Dr. A. NAGARAJA RAO

 2nd Derivative _ Laplacian

 f  f 2 2
 f  2  2
2
x y

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

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

 2 f  [ f ( x  1, y )  f ( x  1, y )  f ( x, y  1)  f ( x, y  1)  4 f ( x, y )]

Dr. A. NAGARAJA RAO

Use of 2nd Derivative for Enhancement
Laplacian

Isotropic filter response is independent of the direction of the

discontinuities in the image to which the filter is applied.

 2
f  2
f
 f  2  2
2
x y

Dr. A. NAGARAJA RAO

 Use of 2nd Derivative for Enhancement
Laplacian

1 1 1
1 -8 1
1 1 1

If the center coefficient

of the laplacian mask is
negative

 f ( x, y )   2 f ( x, y )
g ( x, y )  
 f ( x , y )   2
f ( xDr., A.yNAGARAJA
) RAO
Use of 2nd Derivative for Enhancement
Laplacian
 2 f  [ f ( x  1, y)  f ( x  1, y)  f ( x, y  1)  f ( x, y  1)  4 f ( x, y)]
0 1 0
g(x,y) = f(x,y) - 1 -4 1 =
0 1 0

 f ( x, y )
2

Dr. A. NAGARAJA RAO

Un-sharp Masking and High-boost Filtering

Dr. A. NAGARAJA RAO

 Un-sharp Masking and High-boost Filtering

(b)

Dr. A. NAGARAJA RAO

Combining Spatial
Enhancement Methods

Dr. A. NAGARAJA RAO

Combining Spatial
Enhancement Methods

Dr. A. NAGARAJA RAO