Whats wrong with coding

hi , I am trying to implement Image enhancement algorithm Dynamic Quadrant Histogram Equalization Plateau limit

I am trying to implement its first part please have a look on following code whats wrong in this code

 clear;
 clc;

 p=imread('pout.tif');p=p(:,:,1);
  h=imhist(p);

[m,n]=size(p);
Pic=zeros(m,n);
N=m*n;
max1=double(max(p(:)));    
min1=double(min(p(:)));
aa=p;
 ch=cumsum(h);
 m0=min1;
 m1=floor(0.25*(N));
 m2=floor(0.5*(N));
 m3=floor(0.75*(N));
 m4=max1;
 m=[m0 m1 m2 m3 m4];
  L=256;
  n0=0;
  n1=m2*((m1-m0)/(m2-m0));
  n2=m2;
  n3=((L-1-m2)*((m3-m2)/(m4-m2)))+m2;
  n4=L-1;
   for j=0:4
       a=sum(h(m(j):m(j+1)));
       P(j)=a./(m(j+1)-m(j));
       P(j)=h(h>P(j)); % clipped Histogram
       M(j)=sum(P(m(j):m(j+1))); % total of Clipped histogram
   Y(aa==j)=n(j+1)+(n(j+1)-n(j))*(P(j)./M(j)); %Transform function
   end
imshow(Y)

following is the link to paper

http://ieeexplore.ieee.org/xpl/articleDetails.jsp?tp=&arnumber=5681139&queryText%253DAdaptive+Contrast+Enhancement+Methods+with+Brightness+Preserving

[Walter Roberson]

How does the output you get differ from your expectation? Are you receiving an error message? What does pout.tif look like?

[Muhammad Ali Qadar]

I am receiving error

Subscript indices must either be real positive integers or logicals.

   Error in DQHE (line 32)
       a=sum(h(m(jj):m(jj+1)));
Pout.tif should be somewhat enhanced than from original image in brightness and contrast. And actually I am not receiving output but error.

Answer 1

[David Sanchez]

In your code

for j=0:4
       a=sum(h(m(j):m(j+1)));
...
...

the index starts at j=0, which is not matlab's way of handling an array. It should start in 1.

m(j) for j=0, will return an error since m(0) (the 0th element of the array) does not exist.

[Muhammad Ali Qadar]

for jj=1:5
     a=sum(h(m(jj):m(jj+1)));

 ...
 ..

I did the above statement but still the error persists, do you think is there anyother problem

[Image Analyst]

You will find out after you look at this. Basically it's saying that m is negative, zero, or have some fractional part and that it's not strictly integers like it should be. I don't know why - I didn't run the code - but you will after you look at the link I gave and step through your code line by line and examine variables.

[Muhammad Ali Qadar]

Thanks for The video, I try it but I can not figure it out

m=[25,15500,31000,46500,255] 
for jj=1:length(m)
a=sum(h(m(jj:jj+1));

infact following equation in the paper

here m is the above matrix and h is the input histogram if I debug it is like that I am accessing element that is out of bounds e.g. h(53334) where numel(h)=255. So How can I solve this issue please see where I am wrong

Answer 2

[Image Analyst]

OK, that's a different error than you originally had. If h is the histogram, then h has only 256 bins (elements). So why are you trying to set m=[25,15500,31000,46500,255]???? There is no 15500 bins so why are you trying to access it?

Try this:

m0=min1;
m1=floor(0.25*(max1-min1)+min1);
m2=floor(0.5*(max1-min1)+min1);
m3=floor(0.75*(max1-min1)+min1);
m4=max1;

[Muhammad Ali Qadar]

great Analyst,

You got my problem, So infact in paper there is equation that separate the histogram into 4 parts, with following equation

and my problem is that m1 m2 m2 are so big values because of following code

   N=m*n;
   m1=floor(0.25*(N)));
   m2=floor(0.5*(N));
   m3=floor(0.75*(N));

I need to normalize these values first between 0-255 then may be my problem could be shaped. Still need your sugesstions

[Image Analyst]

Try this:

m0=min1;
m1=floor(0.25*(max1-min1)+min1);
m2=floor(0.5*(max1-min1)+min1);
m3=floor(0.75*(max1-min1)+min1);
m4=max1;

[Muhammad Ali Qadar]

Analyst, can also see this that am I writing wrong or right here for the following

first Equation is for clipping the histogram, second is the transform function, and third is the total of clipped histogram

 for jj=1:5     
         a=a+h(m(jj));
        P(jj)=a./(m(jj));
        h(h(jj)>P(jj))=P(jj); % Clipped Histogram
        M(jj)=sum(h(h(jj)>P(jj)));%total of Clipped histogram
        aa(p==jj)=n(jj)+(n(jj)-n(jj))*(P(jj)./M(jj)); %Transform function
       end
following images are resultant if I do it with above coding I think Something Still wrong with my code first one Original and Last Equalized with algorithm,

However accoding paper result should be like this