Fast and efficient MRF-based blotch detection algorithm for degraded film sequences

This paper proposes an efficient blotch detection algorithm based on a Markov Random Field (MRF) model with less computational complexity and with lower false alarm rate than the existing MRF-based blotch detection algorithms. In our algorithm, by restricting the attention of the detection process only to the candidate blotch detection areas, our proposed algorithm can save the computational time for detecting the blotches. The problem of the confusion with the blotches is frequently seen in the vicinity of the moving object due to the poorly estimated motion vectors. To solve the problem, we first find the moving edge pixels using the significant test between the forward and backward motion-compensated frames, and then incorporate the weighting function with respect to the detected moving edge pixels into the MRF model. Experimental result shows that our proposed method provides the computational simplicity and an efficient detecting performance for the blotches.