Eliminating False Positives in Edge Detection/Pavement Image Crack Detection

I have images of pavement taken using a UAV, and am trying to detect sealed and un-sealed cracks in the pavement images. I am having an issue with false positives around the white lines in the pavement, because there are also black lines painted on the pavement next to the white lines which appear similar to the sealed cracks in the image. Is there some way by which I could eliminate these false positives using the knowledge that they are near the white lines in the image? I have thought of trying to find all the black lines near the while lines in the image before processing with the crack deteciton algoirthm (which is a slightly modified version of CrackIt found here: http://amalia.img.lx.it.pt/CrackIT/), and setting them equal to the mean value in the image or to the same value as the white lines, but I have been unable to implement this successfully/efficiently.