Progressive videotext regions inpainting based on edge detection and statistic method

With the rapid speed of spreading video content, there is some videotext in the video content. Usually some videotext is the advertisements and it is not needed for some people. Most image inpainting methods deal with the broken regions occurring when the original image is broken. Few methods can handle the inpainting of the original region occupied by motion videotext well since the complex background of motion videotext. In this paper, we propose a progressive method to recover the original region occupied by the scrolling videotexts. Besides, we utilize the characteristic of the edge along with the videotext region's boundary to form a searching window or use the mean value of neighboring blocks for recovering. The block searching recovery method is carried out by shifting the reference block spirally to compare with the neighboring windows. We have completed the whole system and the experimental results show all of the horizontal and vertical scrolling videotexts can be recovered well.