Video Copy Detecting System Based On Browser/Sever Architecture

With the rapid development of multimedia and Internet, the number of web videos has exploded. Tracing the video content and controlling the copyright of the huge number of videos uploaded is critical challenge for the owner of the popular video web servers.A video copy detection system by example is presented in this paper. Users can upload a query example to the web server, and the similar video clips are queried from the dataset of video clips. These processes can be seen as content-based video copy detection. The theory and structure of the video copy detection system based on Browser/Server architecture is studied. The total video copy detection system comprise four elements:uploading of video file, reading and encoding of video file, key frame detection, extraction and indexing of global video features. First, we implement the uploading of video file using JSPSmartUpload. Then, the video files of AVI are read and encoded using DirectShow technique. The key frames can be detected using the color histogram method. The global color features are extracted from the key frames, and the video feature is a set of color histograms. The Bounded Coordinate System method is used to index the video feature. The experiment results show that our system can find the video copies from a large dataset of 5,000 videos extremely fast.