Digital Image And Video Inpainting

Digital image and video inpainting technology is to use the neighborhood information of the area to be patched or the information of the before and after frames in dynamic image to estimate the defect information of the area to be remedied, which can reach the visual effects before damaged.Digital image and video inpainting technology is used widely. Such as in a scene where we need to hide a target, remove a specific object or wipe off some words in photos or videos, or restore the flicker, mosaic, spots or scratches in the damaged films, as well as the removal of the damage caused by the creases and stains of the lens when shooting and so on. In recent years, images and video inpainting is concerned increasingly and researched more and more deeply. In this paper we mainly research the image inpainting for the larger-scale damage and video inpainting for mosaics.For the image inpainting, image inpainting based on partial differential equations is better for small-scale damage, and texture-based image inpainting is better for the larger-scale damage. Therefore, in this paper we improve the texture-based image inpainting algorithm-CIEI (Cross Isophotes Exemplar-based Inpainting), so that it has better applicability to restore the texture-rich and larger-scale damaged image. And the results show that the algorithm has achieved good results.For the video inpainting, automatic detection and restoration for mosaic and the scratches are our research focus. According to the difference in the brightness characteristics between the scratch and the surrounding scenery, the algorithm can automatically detect the damage location. Depending on the blocky mosaic features appeared in the video, it can also auto-detect the damage location. Then it can repair them using the fusion information in the before and after frames and the image restoration algorithm. Experimental results show that the recovery has achieved good results.