I am working on satellite images to mask out the rooftop area from houses and buildings.
Applying histogram equalization on the image and then applying the watershed algorithm to mask out the buildings, I am getting too many responses and without applying that some rooftops are missing. Is there any way to optimize the number of counts? Here are some sample images that I have to work on. 
I am applying Mean Shift Filtering and after that thresholding and then watershed algorithm.
↧