Temporal Consistency Based Method for Blind Video Deblurring

In this paper, we propose a video deblurring method for recovering effectively the blurred video caused by camera shake. First, we estimate the blur kernel for each blurred video frame by introducing a preprocessing strategy which employs the anisotropic diffusion and shock filter to get strong edges from each video frame. Second, for video deblurring, we propose a temporal cubic rhombic mask technique, which can not only suppress ringing artifacts, but also can enhance temporal consistency of the deblurred video. The proposed technique uses the relationship of the current frame and the frames before the current frame, and the frame behind the current frame for obtaining a high-quality deblurred video. Experiments demonstrate that proposed method outperforms state-of-the-art deblurring methods for the blurred video.