An efficient video signature & bitmap indexing algorithm for video copy detection

In video copy detection, both storage space and computation complexity (which lead to high power consumption) in the searching/matching grows with the vast amount of video content as well as the sizes of the indexes of the video signatures. In this paper, we improved on our previous work by: revision of meta-slice generation; introduction of a two-layer bitmap index and matching algorithm to reduce storage space and computational time. Experimental results show that the proposed system reduces storage space by about 6.4 times and computational time by about 4 times. Despite the reductions, the system maintains its robustness as well as accuracy of 100% precision and 97.6% of recall as the original system.