Ideas for breaking up an image based on pixel values

I'm dealing with greyscale images made up around of 884 rows. Say for example, at the top of the image from rows 0-200, are of a low intensity therefore close to white on the greyscale (200-256). The rows that make up the middle of the image (rows 200-600) are medium on the grey scale (100-200). and finally the rows that make up the bottom of the image are closer to back. That makes up 3 main regions on the image.

The beginning and end of these regions may vary over images (which rows the regions begin and end). Is there any ideas you could share for automatically marking regions based on pixel values as such on an image??

Answer 1

[Siam]

You can have a look at region based segmentation

http://www.mathworks.com/matlabcentral/fileexchange/authors/116449