Research On Digital Video Copy-move Forgery Detection

With the development of modern information technology industry,all kinds of digital multimedia equipment has been widely used in every aspect of our lives in modern society,such as digital videos.As an important kind of multimedia,digital video has been broadly applied to news reports,education,commercial advertisement,sworn evidence and so on.As an result,it has become an significant way for people to acquire knowledge and information.But there are some criminals who want to tamper with videos for certain purposes maliciously,so it is an urgent problem to find the method to distinguish original and forgery videos correctly and effectively.This paper mainly talks about the way to detect intra-frame copy-move tampering of videos.It is a kind of tampering to copy a region of the frame and paste it to another area in the same frame.The detection algorithm of this type of tampering is mainly divided into two kinds.One is based on the extraction and matching of key points,and the other is by means of extraction and matching subblocks' feature.For the reason that the formal method is broadly approved and put into use for its relatively strong robustness,our researches for intra-frame copy-move tampering detection also made use of method based on key points.And our works have been recorded in this paper as follows:(1)For intra-frame homologous copy-move tampering,we propose an algorithm based on CSIFT,which is a fusion of SIFT and color information.CSIFT adds color invariant feature information in step of key points extraction,improving the matching results.Our algorithm first used the SSIM to segment the video frame sequence into several parts and extracted a key frame for each sequence part.Then the key points of these key frames are extracted accurately and efficiently by CSIFT.Finally,we located the copy-move areas and calculated the position of the tampering regions on the subsequent frame by RCT tracking algorithm.Through our experiments,the robustness of the algorithm has been verified,and the time efficiency and accuracy are higher and better than the algorithms based on SIFT and other features.(2)In another section we propose a copy-move forgery detection method based on the improved FREAK feature for the low quality videos.The improved FREAK retained the original retina sampling mode,but in order to make it more adapt to low quality pictures and videos,we adjusted the concentric layers,size of the receptive field,and dimension of the feature vector.The experiment shows that the improved FREAK is a good descriptor and combined with the overall detection algorithm in this paper,it can accurately detect the copy-move tampering in low-quality videos.The experiment also proves that this algorithm has reasonable detection result for high quality video.