Adaptive Threshold for Segmentation

I have a piece of code which works perfectly fine but i am not able to understand the part where it creates a threshold matrix. Can someone please take time out and explain it to me ?

% A1 is the input image

A1max=ordfilt2(A1,512*512,ones(512,512)); %my max value % 512x512 Maximum Filter

A1min=ordfilt2(A1,1,ones(512,512)); %mIN value % 512 X 512 Minimum Filter

A1dif=A1max-A1min;

avg=filter2(fspecial('average',7),A1)';

z=0.9;%%GAMMA

c=0.5;%%ALPHA

[ka, kb]=size(A1);

for i=1:ka

    for j=1:kb

              if  avg(i,j)> c*A1dif(i,j) && (abs(avg(i,j)-A1(i,j))/(avg(i,j)))<(1-c)   

                       if avg(i,j)>=A1(i,j)
                                 TH(i,j)=c*avg(i,j);

                       else
                                TH(i,j)=avg(i,j);

                      end
              else
                               TH(i,j)= avg(i,j)+z*A1dif(i,j);        
             end    
    end
end

How does one pick the Gamma and Alpha values?

Answer 1

[Image Analyst]

I'd say it's just by trial and error to see what looks good. You might want to use rangefilt() instead of doing ordfilt twice. Also, you vectorize the loops to speed it up and make it more compact.

Man, that's one weird filter. What's its purpose?

clc;
clear all;
clc;    % Clear the command window.
close all;  % Close all figures (except those of imtool.)
clear;  % Erase all existing variables. Or clearvars if you want.
workspace;  % Make sure the workspace panel is showing.
format long g;
format compact;
fontSize = 20;

% A1 is the input image
A1 = imread('cameraman.tif');
% Display the original gray scale image.
subplot(1, 2, 1);
imshow(A1, []);
title('Original Grayscale Image', 'FontSize', fontSize);
% Enlarge figure to full screen.
set(gcf, 'Units', 'Normalized', 'OuterPosition', [0 0 1 1]);
% Give a name to the title bar.
set(gcf, 'Name', 'Demo by ImageAnalyst', 'NumberTitle', 'Off') 

A1max=ordfilt2(A1,512*512,ones(512,512)); %my max value % 512x512 Maximum Filter
A1min=ordfilt2(A1,1,ones(512,512)); %mIN value % 512 X 512 Minimum Filter
A1dif=A1max-A1min;
avg=filter2(fspecial('average',7),A1)';
z=0.9;%%GAMMA
c=0.5;%%ALPHA
[ka, kb]=size(A1);
for i=1:ka
  for j=1:kb
    if  avg(i,j)> c*A1dif(i,j) && (abs(avg(i,j)-A1(i,j))/(avg(i,j)))<(1-c)
      if avg(i,j)>=A1(i,j)
        TH(i,j)=c*avg(i,j);
      else
        TH(i,j)=avg(i,j);
      end
    else
      TH(i,j)= avg(i,j)+z*A1dif(i,j);
    end
  end
end
subplot(1, 2, 2);
imshow(TH, []);
title('TH Image', 'FontSize', fontSize);

It looks like it filters twice, once on the transpose of the image and adds them together. What is this for?

[nazneen]

Can you tell me what you mean by vectoring the loop? And I have a problem understanding the code itself. I don't understand why I am using alpha and gamma in the first place and what the arguments of the comparators denote.

[Image Analyst]

The vectorized code is attached below the image.

[Rakshith Nayak]

how did you get alpha ,gamma and 'average' 7(line 23) values?

[Image Analyst]

Personally, I don't know, and nazneen hasn't been seen here in 5 years.