Image Inpainting Via Video Frames

Image inpainting technique is the process to reconstruct or filling-in missing or damage parts of image or video in a way that is undetectable to the casual observer. We have specifically address the problem of de fencing of image from fenced image or video as there are fences or barricades at public places due to various reasons like security, which hindered our capability to capture scene of interest. Our proposed approach consists of three steps; initially we will detect a fence using image matting approach. After detecting the fence we will determine the relative motion of pixel, in our approach we have exploit the fact that region invisible in one frame might be visible in other frame and we have model image to be inpainted or frame as Markov Random Field for efficient repainting. The last step in our approach is inpainting of missing (fence covered) region using primal dual total variation algorithm. In this paper we have provided brief overview of various task and techniques use for image inpainting.