A study of image completion for reducing filling errors
Gang Wu, Qiang Wei · 2010
Image completion aims at inferring contents in the missing region of an image. Exemplar-based methods provide a powerful tool to solve this problem, but suffer from filling errors in some situation. We uncover one important reason causing the filling errors and propose a solution by making use of the estimated structure intersection to adjust the filling priority. Experimental results verify the performance of the proposed method.