Wondering if imagemagick or graphicsmagick would be useful. They can be used as libraries instead of just programs. libmupdf has some interesting graphics routines. As to boundary detection, might need to look at machine learning libraries for that. Some C machine learning libraries that deal with image detection include ccv ( https://github.com/liuliu/ccv ), yolo ( https://pjreddie.com/darknet/ ) and vlfeat ( https://www.vlfeat.org/ ).