Scribd
Upload
32 views

6 - Image Segmentation - Unit 3

The document discusses image segmentation techniques in digital image processing. It covers key stages in image processing like image segmentation. It describes different approaches to image segmentation including discontinuity based (using edge detection) and similarity based (using thresholding, region growing, splitting and merging). Specific techniques covered are linking edge points using local and global/Hough transform processing, and thresholding methods like local, global, adaptive thresholding. Region based methods and their algorithms are also summarized.

Uploaded by

Abysz
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
32 views

6 - Image Segmentation - Unit 3

The document discusses image segmentation techniques in digital image processing. It covers key stages in image processing like image segmentation. It describes different approaches to image segmentation including discontinuity based (using edge detection) and similarity based (using thresholding, region growing, splitting and merging). Specific techniques covered are linking edge points using local and global/Hough transform processing, and thresholding methods like local, global, adaptive thresholding. Region based methods and their algorithms are also summarized.

Uploaded by

Abysz
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 55

“Digital Image Processing”

Course Code- CSE 314 M

Image Segmentation

(Chapter 10 of textbook)

1
Recall…
Key Stages in DIP: Image Segmentation

Ima ge I ma g e
Re s tora tion S egmenta tion

Ima ge Morphologica l
Enha nce me nt P roce s s ing

Ima ge Repre s enta tion


Acquis ition & De s cription

P roble m Doma in Object


Recognition
Colour Ima ge Ima ge
P roce s s ing Compre s s ion
2
On completion we will learn
• What is image analysis, image segmentation and its role?

• Different approaches to image segmentation


– Discontinuity based
– Similarity based

• Discontinuity (point, line, edge) detection techniques


➢ Linking of edge point using Local processing
➢ Linking of edge point using Global processing (Hough Transform)

• Similarity based image segmentation techniques


– Thresholding (Local Thresholding, Global Thresholding, Adaptive Thresholding)
– Region growing
– Region splitting and merging 3
4
5
6
Image Segmentation Techniques…

1. Discontinuity (point, line, edge) detection techniques:


➢ Linking of edge point using Local processing
➢ Linking of edge point using Global processing (Hough Transform)

2. Similarity based image segmentation techniques


– Thresholding (Local Thresholding, Global Thresholding, Adaptive Thresholding)
– Region growing
– Region splitting and merging

7
8
9
10
11
12
13
14
( Local Processing)

15
( Local Processing)

16
( Local Processing)

17
19
20
21
( Global Processing- Hough Transform )

• Hough Transforms takes the images created by the edge detection operators.
• Most of the time, the edge map generated by the local edge detection
algorithms is disconnected.
• “Hough Transforms” can be used to connect the disjointed edge points. It is
used to fit the points as plane curves are lines, circles, and parabolas.

23
( Global Processing- Hough Transform )

24
From lines to points
• A lines is a collection of points.
• Managing a collection of points is tougher than managing a single point.
• So the first thing is to learn is how to represent a line as a single point,
without losing any information about it.
• This is done through the m-c space.
From points to lines

• Any line passing through (xa, ya): ya = mxa +c


• Rearranging: c = - xam + ya
• The above is the equation of a line in the mc space.
• So, a point in the xy space is equivalent to a line in the mc space.
How does this help us?

• The Hough transform is all about doing what we just learned: converting
points in the xy space to lines in the mc space.
• In an edge detected image, and for every point that is non black, we
draw lines in the mc place.
• Obviously, some lines will intersect. These intersections mark are the
parameters of the line.
The points 1, 2, 3, and 4 are represented as various lines in the mc space. And the
intersection of these lines is equivalent to the original line.
( Global Processing- Hough Transform )

29
( Global Processing- Hough Transform )

30
Image Segmentation Techniques…
1. Discontinuity (point, line, edge) detection techniques
➢ Linking of edge point using Local processing
➢ Linking of edge point using Global processing (Hough Transform)

2. Similarity based image segmentation techniques


– Thresholding (Local Thresholding, Global Thresholding, Adaptive Thresholding)
– Region growing
– Region splitting and merging

31
a. Gray Scale Image
b. Histogram
c. Thresholed Image
d. Boundary detection

32
33
34
35
36
37
1

38
2

Input Image Histogram Result of Basic Global


Thresholding

39
40
41
42
The calculations for variances (the measure of spread) for a threshold value 3

43
Input Image Result of Otsu’s Thresholding

44
Input Image Result of Otsu’s Thresholding

45
46
47
48
49
50
51
1. Usually the algorithm starts from the initial assumption that the
entire image is a single region
2. It computes the homogeneity criterion to see if it is TRUE.
3. If FALSE, then the square region is split into the four smaller
regions.
4. Repeat step 2 and 3 for each of the sub-regions until no further
splitting is necessary.
5. These small square regions are then merged if they are similar to
give larger irregular regions.

52
Second Split Merge 53
54
55

You might also like