I have a project and I have to implement histogram equalization without using the histeq function anybody have any idea ?
Check out my histogram shaping app:
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.
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)