Improving The Video Shot Boundary Detection Using The HSV Color Space And Image Subsampling

With the improvement of computer performance, The popularity of photographic equipment and mobile terminal, the development of multimedia technology, Internet and mobile networks, the application of the video is more and more widely. The video has already produced by previous team of professionals and development into people can use anytime and anywhere with some portable devices such as mobile phones, cameras; the transmission of video has been spread by the TV in the past, but the video can be realized via the Internet and mobile network to upload and download at any time at present; video is stored by conventional film or tape to digital media. Because of these aspects of development and progress, the number of video is increasing rapidly, the classification, management and retrieval of video is more and more difficult. Content Based Video Retrieval has been a hot research topic in the field of video, its purpose is to help people in huge amounts of video search to their content of interest, and the key technology of Content Based Video Retrieval is shot boundary detection.This paper introduces the research background and the Content Based Video Retrieval technology and the research status of shot boundary detection, shot boundary detection is introduced the related basic theory, summarized and analyzed based on the compressed domain and compressed domain many kinds of shot boundary detection algorithms.This paper puts forward a kind of shot boundary detection method based on HSV color space and image sampling. We use the value of the dimension of V in HSV color space as the characteristic value; the r’th frame of the video is segmented into many blocks with certain size, the we extract the value of V and the space position of the center of block, in the first t+n’th frame in video sequences, we can search a small square region whose length is determined and the center position to be the same as the the point that we select in the previous frame t’,and their value could be the closest; all the minimum value of the mean is the difference between two frames; an improved double thresholds method is proposed for the Cut Transition and Gradual Transition detection; Respectively we build the model the Cut Transition and Gradual Transition detection.Finally we carried on the experiment, the results show that recall and precision of the proposed algorithm is higher than two paper which is compared, so the proposed algorithm is more advanced.