Extraction of Key Frames Using Rough Set Theory for Video Retrieval

Abstract

A key frame is an illustrative frame consisting of the entire shot data. It is used to analyze, classify, index and recover video. The present algorithms produce appropriate representative frames and also produce irrelevant representative frames. Some of the algorithms are not able to generate exact key frames for entire shot. To overcome this problem, we proposed a better and efficient scheme based on DC coefficients and rough sets to achieve better results when compared to the rest. This proposed algorithm extracts the exact key frames since it eliminates the distinctness of the selection of key frames. This algorithm is applicable only for compacted MPEG videos. So decoding is not necessary. Thus, the performance of the proposed algorithm exhibits its efficacy in results.