Research And Implementation Of Video Streaming Content Detection

Digital video has become the major media form in multimedia applications. Alarge number of video applications, led by video monitoring system, gradually appear inpeople’s daily life. As a result of the wide application of digital video, mass video datais generated. The video data is large in quantity and high in compression rate. Quick andeffective inquiry of the mass video data become a problem to be solved.Image frame extraction and image matching are the two key technology of videostreaming content detection. The speed of image frame extraction greatly influences thespeed of the system. For some video streaming, we can choose reasonable strategy ofimage frame extraction to improve the speed. In image matching, we comprehensivelyuse image layering and two-stage strategy. First we use wavelet decompositionapproach to do image layering, and then we do image matching in two stage. In firststage we use gray histogram to do rough matching in low resolution image. In secondstage we use Zernike moment to do fine matching in original image. In rough matching,a threshold is needed to identify qualified candidates. If there is not any qualifiedcandidate, image matching process end. On the basis of a large amount of research onvideo streaming content detection, a video streaming content detection system isimplemented. In this system, we use many types of video to perform a lot of experiment.The experimental result show that the system can quickly detect the content of videostreaming and has high accuracy.