Image Enhancement

By
Dr. Pradyut Kumar Biswal
IIIT, Bhubaneswar
Introduction

➢ Enhancement is the process of manipulating an

image so that the result is more suitable than
the original for a specific application.
➢ Image enhancement is the process of making
images more useful.
➢ The reasons for doing this include:
▪ Highlighting interesting detail in images
▪ Removing noise from images
▪ Making images more visually appealing

9/5/2024 Digital Image Processing 2

Image Enhancement Examples

9/5/2024 Digital Image Processing 3

Image Enhancement Examples (cont…)

9/5/2024 Digital Image Processing 4

Image Enhancement Examples (cont…)

9/5/2024 Digital Image Processing 5

Image Enhancement Examples (cont…)

9/5/2024 Digital Image Processing 6

Domains

➢ There are two broad categories of image

enhancement techniques

❑ Spatial domain techniques

❖ Direct manipulation of image pixels

❑ Frequency domain techniques

❖ Manipulation of Fourier transform or
wavelet transform of an image

9/5/2024 Digital Image Processing 7

Spatial Domain

➢ Various techniques of image enhancement

are:
▪ Point processing
▪ Mask Processing
▪ Histogram based

9/5/2024 Digital Image Processing 8

Point Processing

➢ Image enhancement using point

processing techniques:
▪ Negative images
▪ Increasing or decreasing brightness
▪ Thresholding
▪ Logarithmic transformation
▪ Power law transforms
▪ Grey level slicing
▪ Bit plane slicing

9/5/2024 Digital Image Processing 9

Spatial domain Image Enhancement

➢ Most spatial domain enhancement operations can

be reduced to the form
g (x, y) = T[ f (x, y)]Origin x
➢ where f (x, y) is the
input image, g (x, y) is
the processed image
and T is some
operator defined over (x, y)

some neighbourhood
of (x, y)

y Image f (x, y)

9/5/2024 Digital Image Processing 10

Intensity Transformation

➢ When g depends only on the value of f at a

single point (x,y), T becomes an intensity (also
called a gray-level) transformation function of
the form

➢ Here, s and r denote the intensity of g and f at

any point (x,y) respectively.
➢ These point operations are also called zero
memory operations.

9/5/2024 Digital Image Processing 11

 Negative Images
➢ Negative images are useful for enhancing
white or grey detail embedded in dark
regions of an image
▪ Note how much clearer the tissue is in the
negative image of the mammogram

Original Negative
s = 1.0 - r
Image Image

9/5/2024 Digital Image Processing 12

Negative Images

Original Image Enhanced Image x

x

y Image f (x, y) y Image f (x, y)

s = intensitymax – r
S = L -1 - r
9/5/2024 Digital Image Processing 13
Brightness modification
s=r±k Where k is a constant.

Original Image

Increase in Brightness Decrease in Brightness

9/5/2024 Digital Image Processing 14

Brightness modification

Original Image

Increase in Brightness Decrease in Brightness

9/5/2024 Digital Image Processing 15

Thresholding
➢ Thresholding transformations are
particularly useful for segmentation in
which we want to isolate an object of
interest from a background
s
255

128 r

1.0 r > threshold

s=
0.0 r <= threshold

9/5/2024 Digital Image Processing 16

Intensity transformation

Contrast Stretching Thresholding

9/5/2024 Digital Image Processing 17

Basic Gray Level Transformations

➢ There are many different kinds of grey

level transformations
➢ Three of the most
common are:
◼ Linear
 Negative/Identity
◼ Logarithmic
 Log/Inverse log
◼ Power law
 nth power/nth root

9/5/2024 Digital Image Processing 18

Logarithmic Transformations

➢ The general form of the log transformation

is s = c * log(1 + r)
where usually c is set to 1.
➢ The log transformation maps a narrow
range of low input grey level values into a
wider range of output values

➢ The inverse log transformation performs

the opposite transformation

9/5/2024 Digital Image Processing 19

Logarithmic Transformations

9/5/2024 Digital Image Processing 20

Power law transformations

➢ Power law transformations have the

following form
s=c*rγ
➢ Map a narrow range
of dark input values
into a wider range of
output values or vice
versa
➢ Varying  gives a whole
family of curves
9/5/2024 Digital Image Processing 21
Power law transformations

➢ The images show a

magnetic resonance s = r 0.6
(MR) image of a
fractured human
spine

s = r 0.4
➢ Different curves
highlight different
detail

9/5/2024 Digital Image Processing 22

Piecewise Linear Transformation

➢ Rather than using a well defined

mathematical function arbitrary user-
defined transforms can be used.
➢ Contrast stretching linear transform is used
to add contrast to a poor quality image.

9/5/2024 Digital Image Processing 23

Piecewise Linear Transformation

➢ Contrast stretching linear transform can be

expressed as:
s =  r, 0 ≤ r ≤ A
= (r-A) + vA A≤ r ≤ B
=  (r-B) + vB B≤ r ≤ L-1

9/5/2024 Digital Image Processing 24

Gray level slicing
➢ Highlights a specific range of grey levels
▪ Similar to thresholding
▪ Other levels can be suppressed or maintained
▪ Useful for highlighting features in an image

9/5/2024 Digital Image Processing 25

Gray level slicing

With Background
Without Background

S = L-1, A ≤ r ≤ B S = L-1, A ≤ r ≤ B
= 0, Otherwise = r, Otherwise

9/5/2024 Digital Image Processing 26

Bit Plane Extraction / Bit Slicing
➢ Often by isolating particular bits of the
pixel values in an image we can highlight
interesting aspects of that image
➢ Higher-order bits usually contain most of the
significant visual information
➢ Lower-order bits contain subtle details

9/5/2024 Digital Image Processing 27

Bit Plane Extraction / Bit Slicing

9/5/2024 Digital Image Processing 28

Bit Plane Extraction / Bit Slicing

[10000000] [01000000]

[00100000] [00001000]

[00000100] [00000001]

9/5/2024 Digital Image Processing 29

Bit Plane Extraction / Bit Slicing

9/5/2024 Digital Image Processing 30

Bit Plane Extraction / Bit Slicing
Reconstructed image
using only bit planes 8
and 7

Reconstructed image
using only bit planes 8, 7
and 6

Reconstructed image
using only bit planes 8, 7,
6 and 5
9/5/2024 Digital Image Processing 31
Spatial Filtering

➢ The principal approach for defining spatial

neighborhoods about a point (x, y) is to use a
square or rectangular region centered at (x, y).

9/5/2024 Digital Image Processing 32

Importance of Neighborhood

➢ Both zebras and dalmatians have black and

white pixels in similar numbers.
➢ The difference between the two is the
characteristic appearance of small group of
pixels rather than individual pixel values.
9/5/2024 Digital Image Processing 33
Spatial Filtering

➢ A spatial filter consists of (a) a

neighborhood, and (b) a predefined
operation
➢ Linear spatial filtering of an image of size
MxN with a filter of size mxn is given by
the expression

9/5/2024 Digital Image Processing 34

Spatial Filtering

9/5/2024 Digital Image Processing 35

 The Spatial Filtering Process
Origin x
a b c j k l
d
g
e
h
f
i
* m
p
n
q
o
r
Original Filter (w)
Simple 3*3 Image Pixels
e 3*3 Filter
Neighbourhood

eprocessed = n*e + j*a + k*b +

l*c + m*d + o*f + p*g + q*h +
r*i
y Image f (x, y)

The above is repeated for every pixel in the original

image to generate the filtered image
9/5/2024 Digital Image Processing 36
Spatial Filtering (Moving Average)

9/5/2024 Digital Image Processing 37

Spatial Filtering (Moving Average)

9/5/2024 Digital Image Processing 38

Spatial Filtering (Moving Average)

9/5/2024 Digital Image Processing 39

Spatial Filtering (Moving Average)

9/5/2024 Digital Image Processing 40

Spatial Filtering (Moving Average)

9/5/2024 Digital Image Processing 41

Spatial Filtering (Moving Average)

9/5/2024 Digital Image Processing 42

Spatial filters

1.Smoothing Spatial filters [low pass].

2.Sharpening Spatial Filters[high pass].

9/5/2024 Digital Image Processing 43

Spatial filters : Smoothing

➢ The low pass filter preserves the smooth

region in the image and removes sharp
variations leading to blurring effect.
➢ Use: for blurring and noise reduction.

Type of smoothing filters:

1. Standard average linear
2. weighted average.
Order statistics
3. Median filter

9/5/2024 Digital Image Processing 44

Smoothing Spatial Filters
➢ One of the simplest spatial filtering operations
we can perform is a smoothing operation
◼ Simply average all of the pixels in a neighbourhood
around a central value
1/ 1/ 1/
◼ Especially useful 9 9 9 Simple
in removing noise averaging
1/ 1/ 1/
from images 9 9 9 filter
◼ Also useful for 1/ 1/ 1/
highlighting gross 9 9 9
detail
◼ An averaging filter in which all coefficients are equal
also called as box filter.

9/5/2024 Digital Image Processing 45

 Smoothing Spatial Filtering
Origin x
104 100 108 1/ 1/ 1/
9 9 9
99 106 98

95 90 85
* 1/

1/
9
1/

1/
9
1/

1/
9

9 9 9
1/ 100108
Original Filter
9 /9 /9
104 1 1

Simple 3*3 /9 1106

199 /9 198
/9
3*3 SmoothingImage Pixels
Neighbourhood /9 190
195 /9 185
/9
Filter

e = 1/9*106 + 1/9*104 + 1/9*100 +

1/ *108 + 1/ *99 + 1/ *98 + 1/ *95 +
9 9 9 9
1/ *90 + 1/ *85 = 98.3333
9 9
y Image f (x, y)

The above is repeated for every pixel in the original

image to generate the smoothed image
9/5/2024 Digital Image Processing 46
Smoothing Spatial Filtering

➢ What happens when the Values of the Kernel Fall Outside the
Image??!

9/5/2024 Digital Image Processing 47

First solution :Zero padding

-ve: black border

9/5/2024 Digital Image Processing 48

2nd solution: border padding

9/5/2024 Digital Image Processing 49

 Smoothing Averaging effects

Original image 3 x 3 averaging

5 x 5 averaging 9 x 9 averaging

15 x 15 averaging 35 x 35 averaging

Notice how detail

begins to disappear

9/5/2024 Digital Image Processing 50

Linear smoothing : averaging kernels

weighted average

Used to reduce blurring more.

9/5/2024 Digital Image Processing 51

Spatial Smoothing Linear Filters

➢ The general implementation for filtering an M

× N image with a weighted averaging filter of
size m × n is given

where m=2a+1, n=2b+1.

9/5/2024 Digital Image Processing 52

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,

mid-point filter, alpha-trimmed filter
9/5/2024 Digital Image Processing 53
Median filter

110 120 90 130

becomes
91 94 98 200 95
90 95 99 100

82 96 85 90

Steps:
1. Sort the pixels in ascending order:
90,90, 91, 94, 95, 98, 99, 110, 120
2. replace the original pixel value by the median : 95

9/5/2024 Digital Image Processing 54

Median filter
reduces salt and pepper noise

The original The smoothed The smoothed image

image with salt image using using median
and pepper noise averaging

9/5/2024 Digital Image Processing 55

Median Filtering
➢ Very effective for removing “salt and pepper” noise).

averaging median
filtering

9/5/2024 Digital Image Processing 56

Order Statistic filters (Max)
➢ Median represents 50th percentile of a ranked
set of numbers.
➢ When 100th percentile is considered, is called as
Max filter.

➢ This filter is used to find brightest points in an

image.
➢ Also, as peeper noise has very low values, it
reduces peeper noise.

9/5/2024 Digital Image Processing 57

Order Statistic filters (Min)
➢ When 0th percentile is considered, is called as
Min filter.

➢ This filter is used to find darkest points in an

image.

➢ Also, as salt noise has very high values, it

reduces salt noise.

9/5/2024 Digital Image Processing 58

Order Statistic filters (Midpoint)
➢ The midpoint filter computes the midpoint
between minimum and maximum values in the
area encomposed by the filter.

➢ This filter combines order statistics and

averaging.
➢ It works best for randomly distributed noise like
Gaussian noise or uniform noise.

9/5/2024 Digital Image Processing 59

Order Statistic filters (Alpha Trimmed
mean)

➢ Alpha-trimmed mean filter removes the d/2 highest and

d/2 lowest intensity values. The average of these
remaining mn-d values is called an alpha-trimmed mean
filter.

➢ The value of d can range from 0 to mn-1.

➢ When d=0, this filter reduces to arithmetic mean filter.
➢ When d=mn-1, the filter reduces to median filter.
➢ This is useful for multiple type of noises such as
combination of salt-pepper and Gaussian noise.

9/5/2024 Digital Image Processing 60

Sharpening Spatial Filters (HPF)
➢ The aim is to emphasize the transitions
in intensity.
❖ Laplacian Operator
❖ Unsharp Masking and Highboost Filtering
❖ Using First-Order Derivatives for Nonlinear
Image Sharpening - The Gradient

9/5/2024 Digital Image Processing 61

Sharpening Spatial Filters

➢ The first-order derivative of a one-

dimensional function f(x) is the difference

➢ Any definition for a first derivative:

▪ Must be zero in areas of constant intensity.
▪ Must be nonzero at the onset of an intensity
step or ramp.
▪ Must be nonzero along intensity ramps.

9/5/2024 Digital Image Processing 62

Sharpening Spatial Filters

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

difference

➢ Any definition of a second derivative

▪ Must be zero in areas of constant intensity.
▪ Must be nonzero at the onset and end of an
intensity step or ramp.
▪ Must be zero along intensity ramps.

9/5/2024 Digital Image Processing 63

Sharpening Spatial Filters

9/5/2024 Digital Image Processing 64

Image Sharpening based on First-Order
Derivatives

➢ For function f(x,y), the gradient of f at

coordinates (x,y) is defined as

➢ The magnitude of vector f, denoted as

M(x,y)

9/5/2024 Digital Image Processing 65

Image Sharpening based on First-Order
Derivatives

➢ The magnitude of vector f, denoted as

M(x,y)

9/5/2024 Digital Image Processing 66

Image Sharpening based on First-Order
Derivatives

3×3 region of an
image

Roberts cross-gradient
operators

9/5/2024 Digital Image Processing 67

Image Sharpening based on First-Order
Derivatives

Sobel operators

9/5/2024 Digital Image Processing 68

Sharpening Spatial Filters: Laplace
Operator

➢ The second-order isotropic derivative operator

is the Laplacian for a function (image) f(x,y). It
is defined as

➢ In the x-direction,

➢ In the y-direction,

9/5/2024 Digital Image Processing 69

Sharpening Spatial Filters: Laplace
Operator

➢ The discrete Laplacian of two variables is

➢ The kernel is

Kernel 1

➢ The kernel is isotropic for rotations in increments

of 90° with respect to the x- and y-axes

9/5/2024 Digital Image Processing 70

Sharpening Spatial Filters: Laplace
Operator

➢ The diagonal directions can be incorporated in

the definition of the digital Laplacian by adding
four more terms.
➢ The kernel becomes
Kernel 2

Kernel 3 Kernel 4
➢ Two other Laplacian
kernels are

9/5/2024 Digital Image Processing 71

Sharpening Spatial Filters: Laplace
Operator
➢ Because the Laplacian is a derivative operator,
it highlights sharp intensity transitions in an
image and de-emphasizes regions of slowly
varying intensities.
➢ This will tend to produce images that have
grayish edge lines and other discontinuities,
all superimposed on a dark, featureless
background.
➢ Background features can be recovered while
still preserving the sharpening effect of the
Laplacian by adding the Laplacian image to
the original.
9/5/2024 Digital Image Processing 72
Sharpening Spatial Filters: Laplace
Operator

➢ Image sharpening in the way of using the

Laplacian:

➢ where f(x,y) and g(x,y) are the input and

sharpened images, respectively.

➢ Let c =−1 if the Laplacian kernels center value is

negative and c = 1 if the Laplacian kernels
center value is positive.

9/5/2024 Digital Image Processing 73

Sharpening Spatial Filters: Laplace
Operator

Laplacian image obtained

using the kernel 1
Blurred Image

Image
sharpened Laplacian image
with c =−1 obtained
using the kernel 2 with
c = -1

9/5/2024 Digital Image Processing 74

Unsharp Masking And Highboost Filtering

➢ This technique is being used since 1930s by

the printing and publishing industry to sharpen
images
➢ Subtracting an unsharp (smoothed) version of
an image from the original image is called
unsharp masking.
➢ It consists of the following steps:
1. Blur the original image.
2. Subtract the blurred image from the
original (the resulting difference is called
the mask.)
3. Add the mask to the original.
9/5/2024 Digital Image Processing 75
Unsharp Masking And Highboost Filtering

➢ Let f(x,y) denote the original image and 𝑓(x,y)

ҧ
denote the blurred image.
➢ The mask is given as

➢ Then, add a weighted portion of the mask back

to the original image

▪ When k = 1, it is called unsharp masking.

▪ When k > 1, the process is referred to as highboost
filtering.
▪ Choosing k < 1 reduces the contribution of the unsharp
mask.

9/5/2024 Digital Image Processing 76

Illustration of the unsharp masking

➢ In this 1D intensity profile

example, a vertical ramp
edge shows the
transitions from dark to
light.

➢ The points at which a

change of slope occurs in
the signal are now
emphasized (sharpened).

9/5/2024 Digital Image Processing 77

Illustration of the unsharp masking

Original Image Blurred Image

Masked Image

Unsharp masking Highboost filtering

with k=1 with k=4.5
9/5/2024 Digital Image Processing 78
 Query?
Thanks

9/5/2024 Digital Image Processing 79