Efficient Histogram-Based Indexing for Video Copy Detection

Given a source video sequence, a video copy detec- tion technique tries to find sequences whose contents are identical to, or near-duplicates of, the source. In this paper, we mainly focus on the efficiency issue to propose a novel indexing method for video copy detec- tion. Two indexing algorithms, the min-hashing algo- rithm and the histogram pruning algorithm, are inte- grated based on the histogram's characteristics to speed up the conventional sliding window method. We implement the features of the ordinal signature and the SIFT descriptor individually for evaluation. Experi- ment results show that the proposed method can sub- stantially reduce the computation cost and yield an acceptable detection accuracy rate.