How to find monolayer material regions in microscope images

I have got optical microscope images of material samples on substrate, and I am trying to use image processing toolbox to help detect monolayers in these images (The 1L regions of attached images). I tried to use 'Canny' edge detection function in matlab image processing toolbox with a upper and lower threshold to find the slight color change between 1L regions and background while neglecting sharp edges (like edges between 2L, 3L regions), but this method is not stable and may give fault edges if there are crack or gaps inside many layer regions. I am thinking if a neutral network or other kind of models can be trained to recognize these regions.Any suggestion on this image processing problem?

Answer 1

[Image Analyst]

What defines the 1L regions? Is it the color? The thickness? The aspect ratio? The area? If it's the color, a simple delta E color difference could be good enough. See my demo http://www.mathworks.com/matlabcentral/fileexchange/31118-color-segmentation-by-delta-e-color-difference

[Kittel Le]

Yes the thicker the darker it will be on image. I think this color based demo works well and can generally mark desired regions.