Research Of Character Recognition And Label In Information Process Of News Videos

Multimedia data has been the important information source of information systems nowadays. Faced with so much battlefield multimedia information, existed information systems have disadvantages that information is overloaded and can't be efficiently used. News videos, as public source of information, contain a great deal of important information. It can supply valuable strategic information to decision-makers, save their time and improve decision efficiency by researching and analyzing news videos. Important characters in news videos indicate great current affairs. The realization of recognition and label of important characters plays a crucial role of analysis of news video, but the method of single pattern recognition isn't robust enough and can't meet the needs of information analysis.This paper probes on character recognition and label technology in news video. The main work and contributions of this paper are as follow:1 It has presented a framework of character recognition and label based on pattern recognition and Semantic label in information process of news videos and has discussed some key technologies of it. 2, It has worked over the arithmetic of face detection and recognition and improved on some aspects, such as skin-color detection, the extraction of candidate regions of faces, the design of neural network, all of which support the realization of character recognition and label.3 It has studied the technology of character label based on multi-information fusion in information of news videos, including the association rules of faces and names, assistant association of captions, the analysis of face relation. 4 It has designed and implemented the subsystem of character recognition and label in the system of information process of news videos, including the modules of analysis and query, and analyzed the results of experiments in detail.