DISTRIBUTED VIDEO IDENTIFICATION WITH PERCEPTUAL TAGS IN PEER-TO-PEER NETWORKS

Abstract

In this paper, a distributed solution is proposed for video identification and copy detection in P2P networks, which representsa video file in the network with a set of (64-256) bits, named as perceptual tags. As such information is derived from theperceptual content of the video rather than its bitstream representation as in the case of cryptographic hashes, it provides arobust identification after the alterations in the file names and formats provided that the visual quality of the video is atacceptable levels. The paper first briefly discusses the requirements for a distributed perceptual tagging system consideringthe low computational power and low bandwidth of internet users. Then, it presents the proposed perceptual tag extractionmethod using the temporal differences between the video frame averages and the proposed distributed searching scheme for aP2P implementation. The proposed extraction and searching methods provide robustness to the alterations in video formatsand small additions and cuttings in the video content as the typical processing in P2P environment and also achieve uniformdistribution and storage load between the peers.