Face Close-assisted Video Analysis Of The Detection Technology Research And Implementation

As the most rich media type in various media format and content, video has focused more and more people' s attention. To browse and retrieve video, it is necessary to investigate the structure of video content and the video abstract technology which fusing manifold media feature. Media content includes many clues, such as face, text, audio, etc. This paper probes on the face feature detection technology which assists video analyze. The main work and contribution of this paper is as follow: Investigate the video analyze and abstract technology which based on multi_feature fusion, put forward the viewpoint that face detection technology is important in the assistance of video analyze. Work over the video analyze technology based on face information. Including the video file classification using face and other media information, the indexing of news video based on face, caption and audio. Otherwise, a scene importance degree judgment model which based on face, text, audio and lens' s length is proposed. Lucubrate face detection technology, such as the definition of the face feature which assists video analyze, face detection technology in video stream and single frame video image, designing and improving a face detection algorithm which assists video analyze. A face detection demo system is proposed. The experimental results show that our algorithm is effective and robust.