clear; close all;
I = imread('Sub.png');
I = rgb2gray(I);
figure; imshow(I)
% se = strel('square',3);
% %# Removes uneven lighting and enhances contrast.
% I = imdivide(I,imclose(I,se));
%I2 = imtophat(I, strel('rectangle', [70 15]));
I2 = imtophat(I, strel('square', 45));
figure; imshow(I2)
% Alpha=.047;
% h = fspecial('motion', 10, 5);
% w=gausswin(I2,Alpha) % you'll have to play with N and alpha
% I2 = imfilter(I2,h,'same','symmetric'); % something like these options
level = .047;
BW = im2bw(I2,level);
D = -bwdist(~BW,'chessboard');
D(~BW) = -Inf;
L = watershed(D);
imshow(label2rgb(L,'jet','w'))
The main way to deal with watershed over-segmentation is by designing markers for the objects to be reconstructed. For example, see the following links.
http://fiji.sc/Marker-controlled_Watershed
ftp://ftp.math.ethz.ch/hg/EMIS/journals/HOA/DDNS/Volume2008/384346.pdf
Hi Leena,
What exactly do you want to segment ?
First of all, your image has too many "sharp" elements. Watershed will segment regions following gradients. If your image has many sharp elements, they will act has natural borders during segmentation.
You need to filter your image (gaussian blur and such, probably what your I2 = imfilter(I2,h,'same','symmetric') commented line was intented to do) and make the regions you are interested in more uniform.
Then you can enhance the contrast and try your segmentation again.
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering
- Signal Processing
- Image Processing