Video Copy Detection Based On Temporal Contextual Hashing

With the development of multimedia technology and the explosive growth of video data, content-based video copy detection has attracted considerable attentions in the multimedia and the computer vision community. However, most video copy detection methods only focus on the contents of key frames and ignore their temporal context information. In this paper, we proposed to express the temporal context of key frame as binary codes, and compare the key frames' binary codes by calculating Hamming Distance to achieve temporal verification efficiently and implicitly. Experimental results on the publicly available video database (TRECVID 2009) indicate the proposed approach achieves high efficiency and accuracy.