Geovideo Indexing Method Based On Semantics

GeoVideo is a kind of video data with spatial reference,which has the characteristics of data acquisition easily,rich semantics,spatio-temporal attributes,intuitive expression and so on,and gradually becoming important geographical information source in virtual geographical environment,intelligent city and urban security.With the extensive layout of video surveillance network,the dynamic and sustainable GeoVideo data presents an unprecedented challenge to the rapid retrieval of GeoVideo.How to retrieve the GeoVideo from the massive geographical video data quickly and accurately becomes the key issue to be solved in the geographic spatio video indexing.GeoVideo has rich semantic information,however,the existing spatial indexing method does not take the rich semantics of GeoVideo into account,moreover it is difficult to find the temporal,spatio and semantic relevance of GeoVideo data and can't support the temporal and spatio semantic association of GeoVideo and geographical environment,so it is difficult to realize correlation analysis and deep excavation of geography video.Therefore,it is urgent to develop a new method of geography video index considering semantics.In this paper,the spatio-temporal semantic integration indexing method of GeoVideo is proposed for the efficient indexing of geography video data.The main research contents of this paper include:(1)This paper proposed the spatial and temporal semantic integration index structure,STSR tree.In view of the time,space and semantic association retrieval requirements of GeoVideo data,this paper designs the STSR tree,which integrates the R tree,MAP21 and inverted index structure,and defines the index structure of STSR(Spatio-Temporal Semantic R-tree)tree and the tree node setting in detail.(2)This paper has researched the STSR tree temporal and spatio semantic integration index algorithm.Based on the STSR tree index structure,this paper studies the insertion,deletion and querying algorithms of STSR tree in temporal and spatio semantic integration.For different temporal,spatio and semantic query requirements,the spatio-temporal query,spatial semantic query and spatio-temporal semantic query are realized based on STSR tree.(3)A multi-channel parallel query algorithm based on STSR tree is implemented.According to the parallelism of the tree index,this paper designs the STSR tree multi-channel parallel query algorithm based on the parallel task unit,and realizes the GeoVideo data efficient index of the spatio-temporal semantic integration.Finally,this paper uses the open source geography video data set provided by Duke University to analyze and validate the STSR tree and index algorithm proposed above.In the condition of different spatio temporal semantic query,this paper compared and analyzed the performances of STSR tree,hybrid IR tree and RT tree,and the result showed that the STSR tree works well in query and parallel query speed ratio.So the validity of the STSR tree is demonstrated.