Features Based Spatial and Temporal Blotch Detection for Archive Video Restoration

摘要

In order to restore blotched archive video without causing distortion to areas of the frames that are not corrupted, the locations of the blotches must be identified. Blotch detection usually needs reliable motion estimation to avoid false detection of uncorrupted regions in existing techniques. In this paper a blotch artifacts detection technique, which is based on spatiotemporal blotch image features extraction to avoid the dependence on the motion estimation, is proposed. In order to greatly reduce false alarms due to motion estimation errors the proposed detection technique is first to detect blotch candidates based on their spatial features and then to detect blotches from the blotch candidates by their temporal intensity discontinuities. Experimental results show that the proposed technique significantly improves detection performance and outperforms existing techniques.