Inpainting Algorithm for Video Processing

Video inpainting refers to fill the missing area in video as faithfully as possible. The proposed algorithm is implemented for object removal from video frames, and the missing area is filled with background pixels. The main motive is to fill the missing region which is left out, in an efficient way, so that image looks as natural as possible. One solution to deal with video sequences is to register the frames, so that the missing area can be filled by most appropriate patch from the registered frame. Control point mapping method is used for frame registration. To reduce computational time, proposed method uses registration of frames before starting actual inpainting. After registration, required object is removed from the frame and inpainting is applied on the patch left after removing the object. To inpaint the missing area, method similar to examplar-based inpainting is used. The method can be used for object removal application, reconstruction of video, etc.