Research On Perceptual Hashing Method For Video Copy Detection

With the development of the computer and information,especially the multimedia technology,the digital videos are increasing in the Internet and the problem of information security is getting serious.The content-based video copy detection(CBCD)is an important technique of copyright protection and content management for digital videos.With significant theoretical and application value,it becomes a research hot in multimedia information processing field.The basic thought of CBCD is to measure the feature sequences extracted from the reference video and test video,which can uniquely identify the content of video.This paper reviews the theory and the classical algorithms of video copy detection and perceptual hashing.Based on the existed technologies,this paper focuses on the key-frame extraction and perceptual hashing methods.The main works of the paper are as follow:A method based on PCC and histogram similarity is proposed for key-frame extraction.The boundaries of video shot are depended on the minima of the PCC values and the key-frames are extracted based on the complexity of shot and histogram similarity.Combing the often used video hashing method,it completes the video copy detection.The experiments show that the proposed method can control the number of key-frames for the videos which have different number and complex degree of shots.It shows a good performance in the video copy detection.A perceptual hashing method based on SURF is proposed.The method uses the scale and rotation invariance of SURF and the location of the interest points to generate the video perceptual hashes.The experiments display that it takes advantage of local descriptor and shorten the length of hashes.It has good discriminability and robustness against the attacks of blurring,adding noise,intensity adjusting,lossy compression,rotation-cropping and flipping,and can meet the demand of video copy detection.A video copy detection based on perceptual hashing is designed and implemented.It examines the performance of the two algorithms proposed in this paper.The video copy detection method proposed in this paper not only can extract key-frame exactly,but also generate the perceptual hashing with high distinctiveness and robustness.It will be well suited for the video copy detection applications.