Video copy detection by conducting fast searching of inverted files

Fast content-based video copy detection is challenging because video databases have become extremely large. Conventional video-fingerprint-based copy detection systems that use the inverted-file approach involve many similarity computations based on the Hamming distance. To overcome this problem, a novel fast searching strategy for inverted files is proposed in this paper. The strategy involves simple table look-up and word counting operations for the fingerprint matching process. The similarity of video fragments is based on the number of matched fingerprints among all video candidates. In this method, the offset time is used, and fingerprints are ordered to further select the matched fingerprints from the video candidates. Moreover, a novel regional average fingerprint that is compatible with the proposed fast searching strategy is proposed. An experimental video copy detection system was used with the proposed algorithms, and the proposed algorithms were compared with other state-of-the-art fingerprinting algorithms on TRECVID 2011 dataset for different types of video distortions. In addition, VCDB dataset was also used to demonstrate the accuracy and efficiency of the proposed fast searching strategy while using inverted files to demonstrate the practicality of the method for a large database. The proposed system achieved higher accuracy on VCDB dataset with considerably higher operation speed compared with conventional inverted-file-based searching methods.