An image mosaic block detection method based on fuzzy c-means clustering

Abstract

With the development of the Broadcasting and Video network, the Monitoring System on Digital Video Broadcasting is becoming more and more important. Image recognition technology is widely applied to detect the degraded video in the television observation system. Mosaic block easily occurs in the TV signals, which will degrade the video quality. The conventional mosaic detection algorithm can't distinguish Chinese character from real mosaic block. In order to solve the problem, we propose a mosaic block detection method based on fuzzy c-means clustering to detect mosaic block. The method contains three steps. First, the frame is processed by Sobel edge detection algorithm to obtain the edge image. Second, clustering features are extracted for the fuzzy c-means clustering algorithm. Third, a fuzzy c-means clustering is employed to distinguish mosaic blocks from all the normal image blocks. Experimental results show that our algorithm could differentiate Chinese character with real mosaic block.