How to Normalise an images histogram?

<<

>>

Attached is a histogram of an image I'm doing processing work on. The images are Greyscale, as you can see along the x axis the values go from 0-256. Im looking to create a function that will normalise the 3 tonal regions. Since the tonal spread varies across different images it doesn't work well if I run a function which will give all pixels <50 a value of 25 for example. Is there a method or function/s I could use which can recognise that there will always be 3 dominant pixel values in the different regions and normalise them by removing outliers, hence decreasing pixel variation.

Answer 1

[Image Analyst]

What are the 3 gray level regions? 0 to 100, 101 to 254, and 255? Then what do you want to do? Do you want to change the histogram so that it's totally flat in each of those regions? I think that would give a strange looking image as most histogram equalized images are. But if you want, I have code that lets you map your image to any histogram shape that you want. It's here in my File Exchange http://www.mathworks.com/matlabcentral/fileexchange/28972-custom-shaped-histogram

Or maybe you just want imquantize():

posterizedImage = imquantize(grayImage, numberOfLevels);

If neither of these is what you want, then explain better.