I want to know if we can implement histogram equalization without the histeq function?

I have a project and I have to implement histogram equalization without using the histeq function anybody have any idea ?

Answer 1

[Image Analyst]

Check out my histogram shaping app:

http://www.mathworks.com/matlabcentral/fileexchange/?term=authorid%3A31862

It will give you a perfectly flat histogram, or any other shape you desire. Otherwise you can do the standard, dumb, imperfect method where you use cumsum() but that won't get you a flat histogram, just a stretched-out/redistributed one.

Answer 2

[Sean de Wolski]

FEX:ImageAnalyst's:Demo

Probably worth a read

Answer 3

[Jurgen]

4 bit grayscale example:

     grayLevels  = (0:15); 
        %Histogram of image, counts per level:
     grayHist    = [0 0 40 80 45 11 70 0 0 0 0 0 0 0 0 15];

Make new gray value table with formula:

EQfactor = grayLevels(end)/sum(grayHist);
grayEq   = round(EQfactor*cumsum(grayHist));

Finished, Use LUT for equalizing; apply like 'grayEq(im + 1)'

-

Extra info: Use LUT to find new histogram without image data

newGray = zeros(1,length(grayLevels)); % New histogram of image
% Bit messy now:
values  = grayHist(grayLevels+1);  
newGray(unique(grayEq(grayLevels+1)+1)) = [0 values(values>0)];
stem(newGray);figure;stem(grayHist)

[Musab Abid]

how can we implement it on a image?

[Image Analyst]

@Musab Abid just have grayLevels be your image, like

grayLevels = imread('moon.tif');