Research On Video Inpainting Technology Based On Motion Mask And Edge Fitting

Digital video inpainting is to inpaint and restoring the damaged regions in a video with the information of known region.As human society has entered the information age,digital videos appears in every corner of daily lives.In recent years,digital video inpainting technology has become a research hotspot in computer graphics and image science.It is widely used in the old film restoration,video editing and other related fields.Firstly this thesis introduces the basic theory of digital video inpainting technology,including classical image restoration algorithms and video segmentation technology.Then it analyzes and compares the existing classical digital video inpainting algorithms by experiment and simulation,especially the algorithm based on motion mask.The main work and research results of the paper are as follows:1.In the past video restoration algorithms using exemplar-based inpainting,the exhaustive strategy of searching for the best matching patch makes the computation very cumbersome,and the accumulated matching error easily leads the wrong information spread,a novel fast video inpainting algorithm is proposed.When restoring the damaged background,the information is directly replicated to fill in the damaged region from the constructed panoramic background image;when restoring the damaged foreground,it seeks a best state matching frame by comparing the adjacent running foreground state of the damaged frame,and searches the best matching patch in this frame.Meanwhile,the algorithm improves the way of searching exemplar patch,principle of matching cost and way of confidence term refreshing.Experimental results show that the proposed algorithm is suitable for video with stationary background and periodical running foreground,having better efficiency and restoration effects.2.When a larger damaged area appears in the video screen,the past video inpainting algorithms based on the motion mask usually can not distinguish the boundary between foreground and background in the damaged area,resulting in very poor visual effects of foreground edge.So a new video inpainting algorithm based on foreground edge curve fitting is proposed.The algorithm firstly uses the interactive video segmentation algorithm to extract the edge information of the video accurately,then the damaged edge contour is fitted and reconstructed with the foreground contour information of the adjacent frames.Finally take the restored contour line as a constraint to inpaint the foreground edge pixel band,and the pixels in the remained damaged area are restored by distinguish it belongs to foreground and background.Experimental results show that the proposed algorithm gets better restoration effect as to previous video algorithms in the case that foreground object movement is random and the damaged area is relative large.It also overcomes the problems like the edge of the fracture and the wrong area spread.