Semantic Video Retrieval And Its Application In Concert Video

With the rapid development of the computer network technology,a large number of online digital video has appeared.Large number of multi-media technology such as large online digital video,digital libraries,video-on-demand,distance learning and so on has become increasingly popular.The complexity of Video information retrieval is much bigger than simple text information retrieval.In this era of information explosion,how to effectively organize and index video and retrieval video information from Video Media Library efficiently is badly needed.According to the information in this article concert video features as well as the study of retrieval semantic video search in the concert video search applications has been analyzed.In order to solve the semantic gap problem,this paper focus on the concert video concept of semantic based object model as well as the establishment of the retrieval framework of ontology-based annotation.This paper presented and analyzed key technology about video retrieval i.e.shot segmentation,key frame extraction,feature extraction and video similarity matching all of which is basically used in the research method.Then the video data structure is analyzed,based on which the concert video of the semantic object model is constructed.According to the characteristics of the concert video database concert video domain ontology is used to guide the annotation of the semantic object.In this paper, standard web ontology language OWL is used to describe the data structure of the concert video and two information retrieval technologies viz.query expansion and regular expressions are also used.A semantics annotation retrieval frame based on Concert domain ontology is proposed in which an OSQL query language is designed to support Ontology.This kind of frame can assist the organization and retrieval of large-scale video library and improve Concert video streaming retrieval,as well as their retrieval precision and recall.Since Video retrieval is a calculation in high-dimensional space,enhancing the efficiency of retrieval is the key to high-dimensional space for large computing problems.A single value decomposition method is proposed in this paper to switch the high-dimensional space to low-dimensional space,where low-dimensional space-by-dimensional filter out non-similar data sets,so as to narrow the search scope and to increase the efficiency of retrieval.To further enhance the precision of semantic retrieval and to make it meet the needs of users,the paper designed a more user-friendly interface relevance feedback, relevance feedback techniques is used to make up for the inadequacy of the computer so that a better retrieval results close to semantic-level requirements.Finally,a prototype of the concert video retrieval system is designed and realized. The overall framework has completed models about shot segmentation,static extraction of the characteristics,video retrieval,and others.The retrieval system’s validity has been verified through testing and analysis the precision and recall of the system which can meet demand of more music fans in the concert video retrieval.