Research On Digital Video Passive Forensics Based On Texture Feature And Edge Information

Today the appearance of high-quality imaging equipment and advanced multimedia editing software makes it easier to tamper such multimedia data. The tampered video propagation through the network spread false information and confuse the public. Criminals use this method to destory evidences. These affect the social stability in a certain extent undoubtedly. So the research about detection of video tampering has become a hot topic in the domain of information security. There is no time to delay to study integrity and authenticity of the video.This paper studies digital video blind forensics through content features in video internal. In this paper, the latest development and research about digital image and video blind forensics have been summarized. The paper analysises the advantages and disadvantages of present algorithms. Then aims at two different tampering methods this paper proposes respective algorithms.The majority of our work can be summarized as follows:1. Detection of falsification in time domain.Texture feature is an improtant characteristic to describe image and video content. In consideration of content continuity between interframe, this paper uses the correlation of texture feature to detect frame insertion and frame replacement. And the validity was verified through experimental analysis.2. Detection of falsification in space domain.Video keying such as chroma key, luma key is a common method of video tampering in space domain. This paper analysises the content feature changes after video keying, finds out abnormal edge feature of the tampered area. We detect the tampered area through the abnormal characteristic, and then import the idea of targer tracking to detect all frames. This method improved the detection efficiency greatly. Experimental results show that this algorithm are effective and feasible.