A fingerprinting system for video copy detection

In this paper, we propose a video copy detection system by adopting a video fingerprinting technique as kernel. To achieve fast speed, the video fingerprints are generated from videos uploaded by users, and then are put into a retrieval system which prepares references by using a direct hashing algorithm for fast decision. To demonstrate the superiority of this method, several video fingerprints are used for comparisons. The experimental results indicate that our proposed method can perform well. The best performance that we can obtain is 0.35% for false negative rate and 0.1% for false positive rate, by using 1,295 videos for evaluation.