A Novel Crack Detection Algorithm for Solar Panel Surface Images

The detection of cracks on solar panel surfaces is the most important step during the inspection of solar panel, and it has very important practical significance. Recently some automated crack detection techniques that utilize image processing have been proposed. But these methods costs lots of computation time and didn´t get a high accuracy. Aiming at some problems of the existing algorithm, we proposed a new framework to detect the cracks. First we generate a binary mask based on the edge information to find the potential areas which have cracks, then we run a modified percolation algorithm on these areas to find crack pixels, then the tensor voting algorithm helps us to remove noise pixels similar to cracks. Finally we connect those crack pixels to form some crack contours. A large number of experiments show that, our method acquires high accuracy and more complete crack contours with low computation costs. Compared with previous methods, our method is faster, more stable and more effective.