Research On Complex Network-based Video Network Constructing Measurement

With the rapid development of multimedia technology, all kinds of media data have increased dramatically. However, the video data contains much of information, which has become a kind of fundamental expression for the multimedia information. With the rapid growth of video resources, there is an urgent need for effectively stor-ing, indexing and retrieving of video data. Existing video retrieval methods are mostly text-based and video content-based. These methods cannot understand the videos on semantic level and then cannot provide convenient retrieval for users. In this paper, the video information extract from IMDB is used to analyze videos on semantic level. After defined the video entity and computed the similarity between video entities, the Complex-Network concept is applied to construct the video network which uses video entity and the similarity respectively as the network's node and edge.This paper is based on the research date from the Internet Movie Database (IMDB). By using the architectures of the complex network, we analyzed and re-searched the hot issue and the key technologies, as well as combined with the Internet Movie Database and then constructed the video network, which is based on the com-plex network. This paper has completed the following works:Frist, this paper defined the video entity and constructed the video level model by using the organizational structure of the video information in the Internet Movie Database. The low-level information from the video is casted away to define the video entity on semantic-level. And then, in order to more clearly showing how the organi-zation of the video is stored, the more intuitive reaction video information between the hierarchy, as well as video similarity calculations, the video level model has been proposed.Second, this paper implemented the automatic extraction method to the complex network node. Based on the definition of a video entity, we researched the accessing method from the network information, and then propose a new algorithm for simulta-neous collecting the network text and picture messages by using the Web crawler. Our goal is the targeted analysis for the video information (Image and text), to ensure that the graphic information to synchronize real-time access to video information network.Third, this paper quoted the video similarity compute algorithm to propose the network edge construction algorithm. This can prepared for the build of the video network.Then, the video network's constructing prototype system has been designed and implemented. Based on the video entity extracting algorithm and edge constructing algorithm, the video network's prototype system has been constructed. The system's design ideas, the architecture and the implement measurement have been given.Finally, the paper summarized the researches and the implementations of the au-tomatically build for the video network, and also proposed some new methods and directions for the future research.