Combining areas from different 16bit images into one and keep the greyvalue information

Hello,

I am measuring a light profile with very high dynamic difference with a camera. I take several images, say five, with different integrating times. So one image shows the high intensity part of the profile with short integrating time and the other images show other, low intensity parts of the profile at longer integrating times. This results in similar greyscale values of the images, but I want to keep the intensity information. So I rescale the greyscale values of the shorter-time-images to the corresponding values they would be at the longest integrating time. The images are mostly black with the information being only in one area of them.

Now I want to stitch the images together into one 16bit greyscale image (double class) that represents my intensity profile without loosing any information. Just adding with imadd doesn't work for me because I have overlapping areas which would then have too high values in relation to the neighbouring regions. Using imfuse I have the problem that the output is RGB and uint8, which destroys the value information.

I need to find the overlapping areas and create average values of the related images in these areas. The solution I have so far is: I create BW images and detect the relevant area in each using regionprops(BW, Im, {'Area', 'PixelList', 'PixelValues'}); Then I check if the same pixelpositions of PixelList of one image are also present in the PixelList of the others, count the number of images that have values in this Pixelposition, and after adding all images devide each pixel by the correct number for averaging. But this involves a lot of loops and a lot of processing time. So I wonder if there is a more elegant solution, but I cannot find anything.

Any ideas? Thanks, Stef