COMPUTER GRAPHICS

&
IMAGE PROCESSING
MODULE 5

GRAY LEVEL TRANSFORMATIONS

Namitha Ramachandran
2
Module - 5 (Image Enhancement in Spatial Domain and Image Segmentation)

Basic gray level transformation functions - Log transformations, Power-Law transformations,

Contrast stretching. Histogram equalization. Basics of spatial filtering - Smoothing spatial
filter- Linear and nonlinear filters, and Sharpening spatial filters-Gradient and Laplacian.

Fundamentals of Image Segmentation. Thresholding - Basics of Intensity thresholding and

Global Thresholding. Region based Approach - Region Growing, Region Splitting and
Merging. Edge Detection - Edge Operators- Sobel and Prewitt.

Namitha Ramachandran
Basic gray level transformation functions

Enhancing an image provides better contrast and a more detailed image as compared
to non enhanced image.
The transformation function has been given below
S=T(r)

where r is the pixels of the input image and s is the pixels of the output image. T is a
transformation function that maps each value of r to each value of s.
Image enhancement can be done through gray level transformations
There are three basic gray level transformation.
•Linear
•Log Transformation
•Power – law Transformation Namitha Ramachandran
Linear transformation
First, we will look at the linear transformation. The linear transformation includes
simple identity and negative transformation.

Identity transition
In this transition, each value of the input image is directly mapped to the
other value of the output image.
That results in the same input image and output image.

Namitha Ramachandran
Negative transformation
The second linear transformation is negative transformation, which is an invert of
identity transformation.
In negative transformation, each value of the input image is subtracted from the L-1
and mapped onto the output image.
In this the following transition has been don
s = (L – 1) – r
since the input image of Einstein is an
Input Image 8 bpp image,
so the number of levels in this image
are 256.
Putting 256 in the equation,
we get
s = 255 – r
Namitha Ramachandran
Logarithmic transformations
Logarithmic transformation further contains two types of transformation.
Log transformation and inverse log transformation.

The log transformations can be defined by this formula

s = c log(r + 1).

Where s and r are the pixel values of the output and the input image and c is a
constant.
The value 1 is added to each of the pixel value of the input image because if there is
a pixel intensity of 0 in the image, then log (0) is equal to infinity.

So 1 is added, to make the minimum value at least 1. Namitha Ramachandran

During log transformation, the dark pixels in an image are expanded as compare to
the higher pixel values.
The higher pixel values are kind of compressed in log transformation.
Log transformation produces high contrast images .

Input Image Log Transformed Image

Namitha Ramachandran
Wider Output Range

Region unaffected

Namitha Ramachandran
Narrow Input Range
Inverse log transformation
The inverse log transform is opposite to the log transform.

Higher input pixel value, lower output pixel value.

Low contrast output image.

Namitha Ramachandran
Narrow Output Range

Namitha Ramachandran
Wider Input Range
 Power–Law (Gamma) transformations
Power Law Transformation is of two types of transformation nth
power transformation and nth root transformation.
s=cr^γ
Variation in the value of γ varies the enhancement of the images. C is constant.

Different display devices / monitors have their own gamma correction, that’s why they
display their image at different intensity.

This type of transformation is used for enhancing images for a different types of display
devices. (Gamma Correction)
The gamma of different display devices is different.
For example, Gamma of CRT lies in between 1.8 to 2.5, Namitha Ramachandran
Power-law transformation is similar to log transformation, but for different gamma,
value output will be different contrast images.

Namitha Ramachandran
Namitha Ramachandran
COMPUTER GRAPHICS
&
IMAGE PROCESSING
MODULE 5

CONTRAST STRETCHING
Namitha Ramachandran
18
 Piece-wise Linear Transformation
Piece-wise Linear Transformation is type of gray level
transformation that is used for image enhancement.

It is a spatial domain method. It is used for the manipulation of an

image so that the result is more suitable than the original for a
specific application.

Rather than using a well-defined mathematical function, we can use

arbitrary user-defined transforms
Namitha Ramachandran
Contrast Stretching:
Low contrast images occur often due to improper illumination or non-
linearly or small dynamic range of an imaging sensor. It increases the
dynamic range of grey levels in the image.

Contrast stretching is a process that expands the range of intensity

levels in an image so that it spans the full intensity range of the
recording medium or display device.

Namitha Ramachandran
 If slope=1 ,output image=input image
Slope=z
d
If slope>1 , output image is brighter than
the input image

Slope=y If slope<1, output image is darker than

input image

x .r , 0<= r <=a
c Slope=x
S= y.(r-a) +c, a<r<=b
a b

z.(r-b)+d , b<r<=L-1

Namitha Ramachandran
Intensity Level slicing
This technique is used to highlight a specific range of gray levels in a given image
(thresholding).

Other levels can be suppressed or maintained – Useful for highlighting features

in an image.

Two basic themes are:

One approach is to display a high value for all gray levels in the range of interest
and a low value for all other gray levels.

The second approach, based on the transformation brightens the desired range of
gray levels but preserves gray levels unchanged. Namitha Ramachandran
Namitha Ramachandran
COMPUTER GRAPHICS
&
IMAGE PROCESSING
MODULE 5

Histogram processing
Histogram Equalization
Namitha Ramachandran
26
Histogram Processing
In digital image processing, the histogram is used for the graphical
representation of a digital image.

In a graph, the horizontal axis of the graph is used to represent tonal

variations whereas the vertical axis is used to represent the number of
pixels in that particular pixel.

Namitha Ramachandran
Applications of Histograms
It is used to analyze an image. Properties of an image can be predicted
by the detailed study of the histogram.
The brightness of the image can be adjusted by having the details of
its histogram.
The contrast of the image can be adjusted according to the need by
having details of the x-axis of a histogram.
It is used for image equalization. Gray level intensities are expanded
along the x-axis to produce a high contrast image.
Histograms are used in thresholding as it improves the appearance of
the image.
If we have input and output histogram of an image, we can determine
which type of transformation is applied in the algorithm.Namitha Ramachandran
Histogram Equalisation
Histogram Equalization is a computer image processing technique used
to improve contrast in images.

Histogram equalization is used for equalizing all the pixel values of an

image, so that a uniformly flattened histogram is produced.

Histogram equalization increases the dynamic range of pixel values and

makes an equal count of pixels at each level which produces a flat
histogram with high contrast image.
While stretching a histogram, the shape of histogram remains the same
whereas in Histogram equalization, the shape of histogram changes and
Namitha Ramachandran
it generates only one image.
Histogram equalization of the following image
The maximum value of
1 2 1 1 1
intensity in the image is 5
2 5 3 5 2
F(x,y)= 21=2 (0,1)
2 5 5 5 2
22=4(0,1,2,3)
2 5 3 5 2
23=8(0,1,2,3,4,5,6,7)
1 1 1 2 1

Input image

Step 1

Gray 0 1 2 3 4 5 6 7
levels(rk)
No. of 0 8 8 2 0 7 0 0
pixels(nk)
Namitha Ramachandran
 Gray 0 1 2 3 4 5 6 7
levels(rk)
No. of 0 8 8 2 0 7 0 0
pixels(nk)
Step 2: Histogram of input image

6
No of 5
pixels
(nk) 4
3
2
1
0 1 2 3 4 5 6 7 Namitha Ramachandran
Gray levels(rk)
 Step 3
Gray levels No.of pixels Probability Density Sk Sk HG
(rk) (nk) Function(PDF)(P(rk)=nk/n) (cumulative Distribution x Sk x7 Equalized
Function-CDF) max.gray level levels
0 0 0 0 0 0
1 8 0.32 0.32 2.24 2
2 8 0.32 0.64 4.48 4
3 2 0.08 0.72 5.04 5
4 0 0 0.72 5.04 5
5 7 0.28 1 7 7
6 0 0 1 7 7
7 0 0 1 7 7
Max.gray n=25
level
Namitha Ramachandran
Gray levels No.of pixels Probability Distribution Sk Sk HG
(rk) (nk) Function(PDF)(P(rk)=nk/n (cumulative Distribution x Sk x7 Equalized
Function-CDF) max.gray level levels
0 0 0 0 0 0
1 8 0.32 0.32 2.24 2
2 8 0.32 0.64 4.48 4
3 2 0.08 0.72 5.04 5
4 0 0 0.72 5.04 5
5 7 0.28 1 7 7
6 0 0 1 7 7
7 0 0 1 7 7
Max.gray n=25
level
Namitha Ramachandran
 2 4 2 2 2
1 2 1 1 1
4 7 5 7 4
2 5 3 5 2
4 7 7 7 4
2 5 5 5 2 F(x,y)=
4 7 5 7 4
F(x,y)=
2 5 3 5 2
2 2 2 4 2
1 1 1 2 1
Output image
Input image

Namitha Ramachandran
Gray levels No.of pixels Probability Distribution Sk Sk HG
(rk) (nk) Function(PDF)(P(rk)=nk/n (cumulative Distribution x Sk x7 Equalized
Function-CDF) max.gray level levels
0 0 0 0 0 0
1 8 0.32 0.32 2.24 2
2 8 0.32 0.64 4.48 4
3 2 0.08 0.72 5.04 5
4 0 0 0.72 5.04 5
5 7 0.28 1 7 7
6 0 0 1 7 7
7 0 0 1 7 7
Max.gray n=25
level
Namitha Ramachandran
Step 4
Gray 0 2 4 5 7
Levels
No.of 0 8 8 2 7
pixels

Step 2: Histogram of output image

6
No of 5
pixels
(nk) 4
3
2
1
0 1 2 3 4 5 6 7 Namitha Ramachandran
Gray levels(rk)
Perform histogram equalisation on the following image its gray
level distribution is given

Gray 0 1 2 3 4 5 6 7
Level(rk)
No.of 8 10 10 2 12 16 4 2
pixels (nk)

Namitha Ramachandran
14
Gray 0 1 2 3 4 5 6 7
13 Level(r
12 k)
No.of 8 10 10 2 12 16 4 2
11 pixels
(nk)
10

5
4
3
2
1
0 1 2 3 4 5 6 7 Namitha Ramachandran
Gray levels(rk)
Gray levels No.of pixels Probability Distribution Sk Sk HG
(rk) (nk) Function(PDF)(P(rk)=nk/n (cumulative Distribution x Sk x7 Equalized
Function-CDF) max.gray level levels
0 8 0.125 0.125 0.875 1
1 10 0.15625 0.28125 1.96875 2
2 10 0.15625 0.4375 3.0625 3
3 2 0.03125 0.46875 3.28125 3
4 12 0.1875 0.65625 4.59375 5
5 16 0.25 0.90625 6.34375 6
6 4 0.0625 0.96875 6.78125 7
7 2 0.03125 1 7 7
Max.gray n=64
level
Namitha Ramachandran
Gray 1 2 3 5 6 7
Levels
No.of 8 10 12 12 16 6
pixels

Namitha Ramachandran
14
Gray Levels 1 2 3 5 6 7
13
12 No.of pixels 8 10 12 12 16 6
11
10

5
4
3
2
1
0 1 2 3 4 5 6 7 Namitha Ramachandran
Gray levels(rk)
Steps Involved in Histogram Equalisation

1. Identify and list gray levels present in your image.

1 2 1 1 1
2 5 3 5 2 For an image segment like this ,maximum intensity used is 5 here,so
gray levels ranging from 0 to 7 ,because this range include 5 .
2 5 5 5 2
2 5 3 5 2
1 1 1 2 1

2. Identify the frequency of each gray level.

E.g. in the above image gray level 0 frequency is 0 ,gray level 1 frequency is 8, gray level 2 frequency
is 8 and so on.

Namitha Ramachandran
3. Histogram of Input image
Plot the histogram of input image based on gray levels and corresponding frequency
values.
4. Evaluate PDF -Probability Density Function (Normalising in the
range of 0-1)
Use the formula P(rk)=nk/n
Where nk=frequency of gray level , n= maximum gray level value

5. Evaluate CDF -Cumulative Distribution Function for each PDF

value. 𝑘

𝑠 𝑘 = 𝑇 𝑟𝑘 = ෍ Pr 𝑟𝑗 , 𝑘 = 0,1, . . 𝐿 − 1
Namitha Ramachandran
𝑗=0
6. Multiply each cumulative value using the maximum gray value
in the above example gray levels ranging from 0…7, the maximum gray value is 7 (say n)
n xSK

7. Perform round off operation over each transformed value and

get the histogram equalised gray levels .

8. Using the histogram equalized gray levels and the frequency value
plot the histogram of the output image.

9. We can identify the output image segment by mapping

histogram equalised gray levels with initial gray levels .
Namitha Ramachandran
 COMPUTER GRAPHICS
&
IMAGE PROCESSING
MODULE 5

Basics of spatial filtering

(Smoothing Linear Filters)

Namitha Ramachandran
49
Basics of spatial filtering

The spatial domain enhancement is based on pixels in a small range

(neighbor).

This means the transformed intensity is determined by the gray values of

those points within the neighborhood, thus the spatial domain
enhancement is also called neighborhood operation or neighborhood
processing.

If the pixel in the neighborhood is calculated as a linear operation, it is

also called ‘linear spatial domain filtering’, otherwise, it’s called
‘nonlinear spatial domain filtering. Namitha Ramachandran
The process of spatial filtering with a 3 × 3 templates (also known as a
filter, kernel, or window).

Namitha Ramachandran
The response ‘R’ to the template is:

R = w(-1, -1) * f (x-1, y-1) + w(-1, 0) * f (x-1, y) + …+ w( 0, 0) * f (x, y)

+…+ w(1, 0) * f (x+1, y) + w (1, 1) * f( x+1, y+1)

For a filter with a size of (2a+1, 2b+1), the output response

can be calculated with the following function:

Namitha Ramachandran
Smoothing Linear Filters

Image smoothing is a digital image processing technique that reduces and suppresses
image noises, also used for image blurring.

Remove small details from an image prior to object extraction.

Bridge small gaps in lines or curves.

In the spatial domain, neighborhood averaging can generally be used to achieve the
purpose of smoothing.

Commonly seen smoothing filters include average smoothing, Gaussian smoothing,

and adaptive smoothing.

It includes average linear filter and order statistics nonlinear filters.

Namitha Ramachandran
 Average Filters
Also referred to as Lowpass filters.

The idea is to replace the value of every pixel in an image by the average of the
intensity levels in the neighbourhood defined by the filter mask.

The application is noise reduction.

Edges are an important part of an image, applying average filters has undesirable side
effects of blurring an image.

Namitha Ramachandran
Namitha Ramachandran
A spatial average filter in which all the coefficients are equal is known as box filters .

In this weighted average filter, the middle cell is given the

highest weight, which means high priority, this will reduce
blurring during the smoothing process.

Namitha Ramachandran
20 30 50 80 100
Apply standard average filter for pixel at (3,3)
30 20 80 100 100

20 30 60 10 110 Pixel at (3,3) is 60

40 30 70 40 100
1 1 1
50 60 80 30 90
1 1 1

1 1 1

1/9 [20x1+80x1+100x1+30x1+60x1+10x1+30x1+70x1+40x1]

Namitha Ramachandran
 Apply standard average filter for pixel at (1,1)
0 0 0 0 0 0

0 20 30 50 80 100 Pixel at (1,1) is 20

0 30 20 80 100 100
1 1 1
0 20 30 60 10 110
1 1 1
0 40 30 70 40 100
1 1 1
0 50 60 80 30 90

1/9 [0x1+0x1+0x1+0x1+20x1+30x1+0x1+30x1+20x1]

Namitha Ramachandran
 Apply standard average filter for pixel at (2,3)

20 30 50 80 100 Pixel at (2,3) is 80

30 20 80 100 100
1 2 1
20 30 60 10 110
2 4 2
40 30 70 40 100
1 2 1
50 60 80 30 90

1/16 [30x1+50x2+80x1+20x2+80x4+100x2+30x1+60x2+10x1]

Namitha Ramachandran
 Order static(Nonlinear ) smoothing Filters
The response of the filter depends on the ordering or ranking of the pixels in the image.
Replacing the center pixel value with the value determined by the ranking result.

1. Median Filter
The filter replaces the value of a pixel in the image by the median of the gray levels
encompassed by the filter.
Provide excellent noise reduction compared to linear smoothing filters.
Cause less blurring in the image.
Most effective for removing impulse noise (salt and pepper noise).

Namitha Ramachandran
Namitha Ramachandran
 Max Filter
To find the brightest points in an image.
Finds the maximum value in the area encompassed by the filter.
Reduces the pepper noise as a result of the max operation.

Namitha Ramachandran
Namitha Ramachandran
 MIN Filter
To find the darkest points in an image.
Finds the minimum value in the area encompassed by the filter.
Reduces the salt noise as a result of the min operation.

Namitha Ramachandran
 COMPUTER GRAPHICS
&
IMAGE PROCESSING
MODULE 5

Basics of spatial filtering

Sharpening spatial filters-Gradient and Laplacian
Namitha Ramachandran
66
 Sharpening spatial filters-Gradient and Laplacian

Sharpening filters highlight the details of an image.

Applications range from electronic printing, medical imaging, industrial inspection, and
autonomous guidance in the military system.

Namitha Ramachandran
Namitha Ramachandran
 𝝏𝒇
=𝒇 𝒚+𝟏 −𝒇 𝒚
𝛛𝒚

𝝏𝟐𝒇 Namitha Ramachandran

= 𝒇 𝒚 + 𝟏 + 𝒇 𝒚 − 𝟏 − 𝟐𝒇(𝒚)
𝛛𝒚𝟐
Namitha Ramachandran
Namitha Ramachandran
 3 sections :
1)Constant intensity
2)Ramp
3)Step

Namitha Ramachandran
Area of constant intensity: both first-order derivatives and second-order derivatives are
zero.
First order derivative nonzero at onset of ramp and step .
The second-order derivative is nonzero at the onset and end of both the ramp and step.
First-order derivative is nonzero and the second is zero along the ramp.

1st derivative detects thick edges while the second derivative detects thin edges.

2nd derivative has a much stronger response at the gray-level step than 1st derivative.

Second-order derivative enhances fine details(thin edges,lines, including noise.)

than that of 1st order derivative.

Namitha Ramachandran
First Derivative r Filters

Namitha Ramachandran
Gradient in diagonal direction

Edge detection

Edge detection

Namitha Ramachandran
Namitha Ramachandran
Namitha Ramachandran
0 -1 0

-1 4 -1

0 -1 0

-1 -1 -1

-1 8 -1

-1 -1 -1 Namitha Ramachandran
 Fundamentals of Image Segmentation.
Image Segmentation is the process by which a digital image is partitioned into
various subgroups (of pixels) called Image Objects, which can reduce the
complexity of the image, and thus analyzing the image becomes simpler.
Similarity Detection (Region Approach)
This fundamental approach relies on detecting similar pixels in an image – based on a
threshold, region growing, region spreading, and region merging so does
classification, which detects similarity based on a pre-defined (known) set of features.

Discontinuity Detection (Boundary Approach)

This is a stark opposite of the similarity detection approach where the algorithm
rather searches for discontinuity. Image Segmentation Algorithms like Edge
Detection, Point Detection, Line Detection follow this approach – where edges get
detected based on various metrics of discontinuity like intensity, etc Namitha Ramachandran
Thresholding - Basics of Intensity thresholding and Global
Thresholding.

Image thresholding is a technique employed to facilitate easy image

segmentation for various image processing tasks.

Simple thresholding technique (Binary Thresholding)

In a simple thresholding technique, a standard threshold value is set and each pixel
value is compared with the threshold value. If the pixel value is less than the
mentioned threshold value then the value is set to 0 or else it is set to the maximum
value.

Namitha Ramachandran
Namitha Ramachandran
Namitha Ramachandran
Global thresholding: T is constant and applicable over the whole image.

Variable/ Local thresholding: T changes over an image. T at a point (x,y)

is a function of the neighborhood of (x,y).

Dynamic / Adaptic thresholding: T changes over an image. T at any

point(x,y) is a function of spatial coordinate (x,y)

Namitha Ramachandran
Basics of Global Thresholding

When the intensity distributions of objects and background pixels are

sufficiently distinct, it is possible to use a single global threshold
applicable over the entire image.

We have algorithms for estimating automatically the threshold value for each image.

Iterative algorithm

Otsu’s method

Namitha Ramachandran
Namitha Ramachandran
Namitha Ramachandran
 COMPUTER GRAPHICS
&
IMAGE PROCESSING
MODULE 5

FUNDAMENTALS OF IMAGE
SEGMENTATION

(Region growing & region splitting and merging)

Namitha Ramachandran
89
 Fundamentals of Image Segmentation.
Image Segmentation is the process by which a digital image is partitioned into
various subgroups (of pixels) called Image Objects, which can reduce the
complexity of the image, and thus analyzing the image becomes simpler.
Similarity Detection (Region Approach)
This fundamental approach relies on detecting similar pixels in an image – based on a
threshold, region growing, region spreading, and region merging so does
classification, which detects similarity based on a pre-defined (known) set of features.

Discontinuity Detection (Boundary Approach)

This is a stark opposite of the similarity detection approach where the algorithm
rather searches for discontinuity. Image Segmentation Algorithms like Edge
Detection, Point Detection, Line Detection follow this approach – where edges get
detected based on various metrics of discontinuity like intensity, etc Namitha Ramachandran
Region-based Segmentation Approach - Region Growing, Region Splitting and
Merging

A region can be classified as a group of connected pixels exhibiting similar properties.

The similarity between pixels can be in terms of intensity, color, etc. In this type of
segmentation, some predefined rules must be obeyed by a pixel to be classified into
similar pixel regions. Region-based segmentation methods are preferred over edge-
based segmentation methods in case of a noisy image. Region-Based techniques are
further classified into 2 types based on the approaches they follow.

Region growing method

Region splitting and merging method

Namitha Ramachandran
 Region Growing Technique

In the case of the Region growing method, we start with some pixel as the seed pixel
and then check the adjacent pixels.
If the adjacent pixels abide by the predefined rules, then that pixel is added to the
region of the seed pixel and the following process continues till there is no similarity
left. This method follows the bottom-up approach.

In case of a region growing, the preferred rule can be set as a threshold.

For example: Consider a seed pixel of 2 in the given image and a threshold value
of 3, if a pixel has a value less than 3 then it will be considered inside the seed
pixel region. Otherwise, it will be considered in another region.
Hence 2 regions are formed in the following image based on a threshold value of 3.
Namitha Ramachandran
Namitha Ramachandran
 Region Splitting and Merging Technique

In Region splitting, the whole image is first taken as a single region. If the region does
not follow the predefined rules, then it is further divided into multiple regions (usually
4 quadrants) and then the predefined rules are carried out on those regions in order to
decide whether to further subdivide or to classify that as a region. The following
process continues till there is no further division of regions required i.e every region
follows the predefined rules.

In Region merging technique, we consider every pixel as an individual region. We

select a region as the seed region to check if adjacent regions are similarly based on
predefined rules. If they are similar, we merge them into a single region and move
ahead in order to build the segmented regions of the whole image.

Namitha Ramachandran
 Usually, first region splitting is done on an image so as to split an image into
maximum regions, and then these regions are merged in order to form a good
segmented image of the original .

Apply region splitting on the following image. Assume the threshold value be<=4.

5 6 6 6 7 7 6 6 Higher value-Lower value>4 then split

R1 R2
6 7 6 7 5 5 4 7

6 6 4 4 3 2 5 6 7-0=7>4 ,split into 4 quadrants

5 4 5 4 2 3 4 6

0 3 2 3 3 2 4 7

R4 0 0 0 0 2 2 5 6
R3
1 1 0 1 0 3 4 4

1 0 1 0 2 3 5 4 Namitha Ramachandran
 Higher value-Lower value>4 then split

R1

7-4=3<=4 ,NO split

5 6 6 6 7 7 6 6
R1 R2
6 7 6 7 5 5 4 7

6 6 4 4 3 2 5 6

5 4 5 4 2 3 4 6

0 3 2 3 3 2 4 7

R4 0 0 0 0 2 2 5 6
R3
1 1 0 1 0 3 4 4

1 0 1 0 2 3 5 4 Namitha Ramachandran
 Higher value-Lower value>4 then split

R2

7-2=5>4 ,split
R21 R22
5 6 6 6 7 7 6 6 R21
R1 7-5=2<=4 ,NO split
6 7 6 7 5 5 4 7 R2
R22
6 6 4 4 3 2 5 6
R24 7-4=3<=4 ,NO split
5 4 5 4 2 3 4 6 R23 R23
0 3 2 3 3 2 4 7 6-4=2<=4 ,NO split

R24
R4 0 0 0 0 2 2 5 6
3-2=1<=4 ,NO split
R3
1 1 0 1 0 3 4 4

1 0 1 0 2 3 5 4 Namitha Ramachandran
 Higher value-Lower value>4 then split

R3

7-0=7>4 ,split
R21 R22
5 6 6 6 7 7 6 6 R31
R1 3-2=1<=4 ,NO split
6 7 6 7 5 5 4 7 R2
R32
6 6 4 4 3 2 5 6
R24 6-4=3<=4 ,NO split
5 4 5 4 2 3 4 6 R23 R33
0 3 2 3 3 2 4 7 5-4=1<=4 ,NO split
R31 R32
R34
R4 0 0 0 0 2 2 5 6
3-0=3<=4 ,NO split
1 1 0 1 0 3 4 4
R33
1 0 1 0 2 3 5 4 Namitha Ramachandran
R34
 Higher value-Lower value>4 then split

R4

3-0=3<=4 ,No split

R21 R22
5 6 6 6 7 7 6 6
R1 R2
6 7 6 7 5 5 4 7

6 6 4 4 3 2 5 6
R24
5 4 5 4 2 3 4 6 R23

0 3 2 3 3 2 4 7 R32
R31
R4 0 0 0 0 2 2 5 6

1 1 0 1 0 3 4 4
R33
1 0 1 0 2 3 5 4 Namitha Ramachandran
R34
Merging Higher value-Lower value>4 then split

Compare Higher value-Lower value>4 then split in both direction

R1 and R21 (MAX-7,MIN-5)
7-5=2<=4,merge
R21 and R1 (MAX-7,MIN-4)
R21 R22 7-4=3<=4 ,MERGE
5 6 6 6 7 7 6 6
R1 R1- R21 and R22 (MAX-7,MIN-4)
6 7 6 7 5 5 4 7 R2
7-4=3<=4 ,MERGE
6 6 4 4 3 2 5 6
R24
5 4 5 4 2 3 4 6 R23 R1- R21 and R22 (MAX-7,MIN-4)
7-4=3<=4 ,MERGE
0 3 2 3 3 2 4 7 R32
R31
R4 0 0 0 0 2 2 5 6

1 1 0 1 0 3 4 4
R33
1 0 1 0 2 3 5 4 Namitha Ramachandran
R34
Merging Higher value-Lower value>4 then split

Compare Higher value-Lower value>4 then split in both direction

R21 R22
5 6 6 6 7 7 6 6
R1 R2
6 7 6 7 5 5 4 7

6 6 4 4 3 2 5 6
R24
5 4 5 4 2 3 4 6 R23

0 3 2 3 3 2 4 7 R32
R31
R4 0 0 0 0 2 2 5 6

1 1 0 1 0 3 4 4
R33
1 0 1 0 2 3 5 4 Namitha Ramachandran
R34
Merging Higher value-Lower value>4 then split

Compare Higher value-Lower value>4 then split in both direction

R21 R22
5 6 6 6 7 7 6 6
R1 R2
6 7 6 7 5 5 4 7

6 6 4 4 3 2 5 6
R24
5 4 5 4 2 3 4 6 R23

0 3 2 3 3 2 4 7 R32
R31
R4 0 0 0 0 2 2 5 6

1 1 0 1 0 3 4 4
R33
1 0 1 0 2 3 5 4 Namitha Ramachandran
R34
Merging Higher value-Lower value>4 then split

Compare Higher value-Lower value>4 then split in both direction

R4, R31, R34, R24 Satisfy merging condition

R21 R22 in either direction, so form a region.
5 6 6 6 7 7 6 6
R1 R2
6 7 6 7 5 5 4 7

6 6 4 4 3 2 5 6
R24
5 4 5 4 2 3 4 6 R23

0 3 2 3 3 2 4 7 R32
R31
R4 0 0 0 0 2 2 5 6

1 1 0 1 0 3 4 4
R33
1 0 1 0 2 3 5 4 Namitha Ramachandran
R34
Merging Higher value-Lower value>4 then split

Repeat the same with all other regions and finally we get 2 regions

R21 R22
5 6 6 6 7 7 6 6
R1 R2
6 7 6 7 5 5 4 7

6 6 4 4 3 2 5 6
R24
5 4 5 4 2 3 4 6 R23

0 3 2 3 3 2 4 7 R32
R31
0 0 0 0 2 2 5 6
R4
1 1 0 1 0 3 4 4
R33
1 0 1 0 2 3 5 4 Namitha Ramachandran
R34
Edge Detection - Edge Operators- Sobel and Prewitt

Edge detection is a technique of image processing used to identify points in a digital

image with discontinuities, simply to say, sharp changes in the image brightness. These
points where the image brightness varies sharply are called the edges (or boundaries)
of the image.

With the help of edges and lines, an object’s structure is known. That is why
extracting the edges is a very important technique in graphics processing and feature
extraction.

Namitha Ramachandran
Sobel Edge Detection Operator

The Sobel edge detection operator extracts all the edges of an image, without
worrying about the directions. The main advantage of the Sobel operator is that it
provides differencing and smoothing effect.
Sobel edge detection operator is implemented as the sum of two directional
edges. And the resulting image is a unidirectional outline in the original image.

Sobel Edge detection operator consists of 3x3 convolution kernels. Gx is a simple

kernel and Gy is rotated by 90°.

These Kernels are applied separately to the input image.

Namitha Ramachandran
Following is the gradient magnitude:

Namitha Ramachandran
Advantages

Simple and time-efficient computation

Very easy at searching for smooth edges

Limitations:
Highly sensitive to noise
Not very accurate in edge detection
Detect with thick and rough edges does not give appropriate results
Diagonal direction points are not preserved always

Namitha Ramachandran
Prewitt operator
This operator is almost similar to the sobel operator. It also detects vertical and
horizontal edges of an image. It is one of the best ways to detect the orientation
and magnitude of an image.

Advantages:

Good performance on detecting vertical and horizontal edges

Best operator to detect the orientation of an image
Namitha Ramachandran
Limitations:

The magnitude of the coefficient is fixed and cannot be changed

Diagonal direction points are not preserved always

Namitha Ramachandran
 Robert's cross operator

Advantages:
Detection of edges and orientation is very easy
Diagonal direction points are preserved
Limitations:
Very sensitive to noise
Not very accurate in edge detection Namitha Ramachandran