Fast and effective blotch identification algorithm in archive film restoration

The conventional blotch identification algorithms would have to detect each pixel in archive film, which are time-consuming. In this work, an edge detection based algorithm is proposed to limit identification area to blotch edges and pixels around blotch edges, which greatly reduces the computational complexity. First, an edge detection algorithm is adopted to obtain edge points of current frame image. Then, partial non-blotch edge points are omitted by proposed approach. Finally, for the remaining edge points as well as the surrounding pixels, an improved local motion compensation-based SROD is proposed to determine whether the nt is a blotch or not. Experimental results show that the proposed approach gives successful identification performances, provides the computational simplicity and outperforms classical algorithms.