| With the development of digital storage technology and multimedia technology, the data storage, management, retrieval and reuse about a flood of audio and video has become a serious issue. The intelligent retrieval technology of Media Asset Management system overcome the shortcomings of traditional text retrieval, and become a research focus. In this paper, the media capital of intelligent retrieval system requirements of the system on the ground floor of the digital image feature extraction, video camera cutting, a key frame extraction and video scenes of (video camera clustering), and based on fuzzy sets of fuzzy clustering analysis.Cutting the video camera, the paper is based on the overall histogram method. In video shot detection on the basis of video key frame extraction through the lens characterized video camera. Common key frame selection algorithm, and some selected key frame the number of fixed video camera can not fully reflect the changes in the content, while there are some key frame selection too many shortcomings. By using information theory, the meaning of the entropy, design, which is based on gray-scale image entropy key frame extraction algorithm and Test shows that the algorithm will be able to overcome the traditional content-based analysis of the key frame extraction algorithm in the key frame extraction too many shortcomings, while the contents of the lens in accordance with the appropriate number of changes to retain key frame.Key frame reflects only the contents of a scene. A single video camera often do not reflect the complete video semantic information, therefore, requires the lens on the basis of a higher level of video unit to establish a unit for the scene said the semantic level video structure.This paper studies the common clustering algorithm, and using fuzzy math theory tools, a design based on fuzzy clustering video semantic analysis algorithms. The algorithm can provide different clustering accuracy of the output. |
PDF Full Text Request