Film blotch removal with a spatiotemporal fuzzy filter based on local image analysis of anisotropic continuity

The film material of old movies is often corrupted by blotches. We present a practical simple easy-implementable method for removing blotches and restoring their missing data. To cope with the motion, instead of estimating an optical flow directly, we take up the robust approach of a local analysis of spatiotemporal anisotropic gray-level continuity. The approach uses simple first-order directional derivatives to select the most smooth direction for each examined pixel, and puts out the incorruption probability that the examined pixel may not be corrupted by blotches. As the restoration filter, we employ a spatiotemporal fuzzy filter whose response is adaptively controlled according to a fuzzy rule with incorruption probability. We design the fuzzy rule in advance by some training method. Computer simulations are presented.