Central Institute Of Technology , Kokrajhar
IMAGE SEGMENTATION
Based on
Global Thresholding &
Gradient based Edge detection
Presented By:
Roshan Adhikari ( Gau-c-12/86)
Tubur Borgoyary (Gau-c-12/L-187)
Under the supervision of
Dr. Pankaj Pratap Singh
Asst. Prof.
CONTENTS:
Introduction
Image Segmentation
Identified issues in image segmentation
Analysis of image segmentation approach
Segmented algorithm
Thresholding based segmentation
Edge detection method
Line Detection
Global and Local thresholding
INTRODUCTION:
In the current scenario, Images have the lots of information
and the major challenge is to segment the relevant
information from the images.
It can be possible by applying the segmentation approaches
in effective manner.
WHAT IS IMAGE SEGMENTATION?
IMAGE SEGMENTATION IS THE PROCESS OF PARTITIONING A DIGITAL IMAGE INTO
MULTIPLE SEGMENTS . THE GOAL OF SEGMENTATION IS TO SIMPLIFY AND/OR CHANGE
THE REPRESENTATION OF AN IMAGE INTO SOMETHING THAT IS MORE MEANINGFUL
AND EASIER TO ANALYZE . IMAGE SEGMENTATION IS TYPICALLY USED TO LOCATE
OBJECTS AND BOUNDARIES (LINES, CURVES, ETC.) IN IMAGES .
Techniques used for Image segmentation:
Thresholding
Edge Detection
Image Thresholding in Matlab:
>> i=imread(‘imagename.jpg’); //Read the image
>> x=rgb2gray(i); // convert RGB to GRAY scale image.
>> figure, imshow(x); //show the converted image
>> figure, imhist(x); //display histogram of the converted image
>> figure, imhist(x,64);
>> x1=histeq(x); // histogram equalization to enhance the contraction an image
>> figure, imshow(x1); //display the equalization image.
>> figure ,imhist(x1); //display histogram of the eqalisation image.
>> t=max(x1(:)); // maximum value for whole matrix
>> h=x1>=t; //Applying thresholding to image x1 to get logical image showing point.
>> imshow(h) // Display the image.
>> tmax=240; //Taking value of ‘t’ as 240
>> h=x1>=tmax;
>> imshow(h); //Show the image.
Original image:
>> i=imread(‘ima.png’); //Read the image
>> x=rgb2gray(i); // convert RGB to GRAY scale image.
>> figure, imshow(x) //show the converted image
>> figure, imhist(x) //display histogram of the converted image
>> x1=histeq(x); // histogram eqalization to enhance the contraction an image
>> figure, imshow(x1); //display the eqalization image
>> figure imhist(x1); //display histogram of the eqalisation image.
>> t=max(x1(:)); // maximum value for whole matrix
>> h=x1>=t; //Applying thresholding to image x1 to get logical image showing point.
>> imshow(h); // Display the image.
>> tmax=200; //Taking value of t as 200
>> h=x1>=tmax;
>> imshow(h); //Show the image
T=200 T=100
Edge detection method:
>>m-=edge(x,’sobel’);
>>Imshow(m);
Edge detection using sobel technique:
m1=edge(x1,’sobel’,’vertical’)
m1=edge(x,’sobel’,’horizontal’);
MAJOR PROBLEMS:
1.SEGMENTATION IN SIMILAR OBJECT:
Due to similar pixel behavior (or approximately ), there may be problem to
distinguish the objects.
2. Boundary detection :
Boundary detection problems occur due to noise (irrelevant pixels). Edge
detection is the common problem in segmentation and also important for
analyzing the boundaries in images. This technique is generally used for
finding the discontinuities in gray level images.
Data used:
The binary and greyscale image data will used in the
proposed techniques for image segmentation
Software used:
MATLAB
REFERENCE:
1. Digital IMAGE PROCESSING by GONZALEZ.
THANKS

Image segmentation

  • 1.
    Central Institute OfTechnology , Kokrajhar IMAGE SEGMENTATION Based on Global Thresholding & Gradient based Edge detection Presented By: Roshan Adhikari ( Gau-c-12/86) Tubur Borgoyary (Gau-c-12/L-187) Under the supervision of Dr. Pankaj Pratap Singh Asst. Prof.
  • 2.
    CONTENTS: Introduction Image Segmentation Identified issuesin image segmentation Analysis of image segmentation approach Segmented algorithm Thresholding based segmentation Edge detection method Line Detection Global and Local thresholding
  • 3.
    INTRODUCTION: In the currentscenario, Images have the lots of information and the major challenge is to segment the relevant information from the images. It can be possible by applying the segmentation approaches in effective manner.
  • 4.
    WHAT IS IMAGESEGMENTATION? IMAGE SEGMENTATION IS THE PROCESS OF PARTITIONING A DIGITAL IMAGE INTO MULTIPLE SEGMENTS . THE GOAL OF SEGMENTATION IS TO SIMPLIFY AND/OR CHANGE THE REPRESENTATION OF AN IMAGE INTO SOMETHING THAT IS MORE MEANINGFUL AND EASIER TO ANALYZE . IMAGE SEGMENTATION IS TYPICALLY USED TO LOCATE OBJECTS AND BOUNDARIES (LINES, CURVES, ETC.) IN IMAGES .
  • 5.
    Techniques used forImage segmentation: Thresholding Edge Detection
  • 6.
    Image Thresholding inMatlab: >> i=imread(‘imagename.jpg’); //Read the image >> x=rgb2gray(i); // convert RGB to GRAY scale image. >> figure, imshow(x); //show the converted image >> figure, imhist(x); //display histogram of the converted image >> figure, imhist(x,64); >> x1=histeq(x); // histogram equalization to enhance the contraction an image >> figure, imshow(x1); //display the equalization image. >> figure ,imhist(x1); //display histogram of the eqalisation image. >> t=max(x1(:)); // maximum value for whole matrix >> h=x1>=t; //Applying thresholding to image x1 to get logical image showing point. >> imshow(h) // Display the image. >> tmax=240; //Taking value of ‘t’ as 240 >> h=x1>=tmax; >> imshow(h); //Show the image.
  • 7.
  • 8.
    >> i=imread(‘ima.png’); //Readthe image >> x=rgb2gray(i); // convert RGB to GRAY scale image. >> figure, imshow(x) //show the converted image
  • 9.
    >> figure, imhist(x)//display histogram of the converted image
  • 10.
    >> x1=histeq(x); //histogram eqalization to enhance the contraction an image >> figure, imshow(x1); //display the eqalization image
  • 11.
    >> figure imhist(x1);//display histogram of the eqalisation image.
  • 12.
    >> t=max(x1(:)); //maximum value for whole matrix >> h=x1>=t; //Applying thresholding to image x1 to get logical image showing point. >> imshow(h); // Display the image.
  • 13.
    >> tmax=200; //Takingvalue of t as 200 >> h=x1>=tmax; >> imshow(h); //Show the image T=200 T=100
  • 14.
  • 15.
  • 16.
  • 17.
    MAJOR PROBLEMS: 1.SEGMENTATION INSIMILAR OBJECT: Due to similar pixel behavior (or approximately ), there may be problem to distinguish the objects. 2. Boundary detection : Boundary detection problems occur due to noise (irrelevant pixels). Edge detection is the common problem in segmentation and also important for analyzing the boundaries in images. This technique is generally used for finding the discontinuities in gray level images.
  • 18.
    Data used: The binaryand greyscale image data will used in the proposed techniques for image segmentation Software used: MATLAB
  • 19.
    REFERENCE: 1. Digital IMAGEPROCESSING by GONZALEZ.
  • 20.