autonomous image segmentation using histogram

Suppose, I have imhist() of an gray scale image which tells me basically the number of pixels containing the similar color inside different regions of the image. I can set top and bottom of threshold for desired region segmentation by looking at the histogram. But I was wondering if this could be done without my intervention. What I mean is that the program should load one image in a folder and perform histogram analysis, then identify the peaks and width of the signal in all significant histogram mode, set (top and bottom) threshold criteria for segmentation and perform segmentation and then follow the process for other images autonomously. We are presuming all the images in this folder are taken almost at the same time, hence the lighting condition and image type and the object of the images are similar.