Font Size: a A A

Research On Data Management Of Geo Video Based On Semantic Correlation

Posted on:2018-09-24Degree:MasterType:Thesis
Country:ChinaCandidate:Z WangFull Text:PDF
GTID:2310330512984780Subject:Engineering
Abstract/Summary:PDF Full Text Request
GeoVideo is a kind of video data with spatial reference.With the extensive deployment of surveillance network,large-scale geographic video data acquisition is possible.At the same time,the rapid development of spatial information technology,computer technology and network communication technology has also provided support for obtaining spatial reference spatial video data.As the important key data of smart city and urban security,Geo Video data is one of the main spatial data types in new video GIS system management.The effective organization of geographic video data is the primary and important basis for determining the availability and efficiency of video GIS systems.Geographic video in the existing GIS system is mainly distributed as independent attribute management and visualization,and lacks the ability of modeling and correlation analysis.It is difficult to support the depth mining and correlation analysis of geography video data.Therefore,this paper oriented to Effective data organization methods,in which research contents mainly includes:(1)This paper proposesed a geographic video presentation model based on ontology.The model defines the description objects of different semantic levels in the video by the structured description framework,and can describe the video content to the degree of behavior semantics and realize the representation of the geographical video in the semantic association mode.In this paper,the process of constructing the semantic ontology model of geographic video is realized by OWL language,and the association reasoning process of semantic objects and spatiotemporal information in geographic video is described based on DL language.(2)This paper studied the method of geographic video relevance measurement.Taking dynamic target of the geographic video as main objects,based on the dynamic target trajectory coordinates and the time dimension,the spatial and temporal correlation model is established.The semantic relevance of GeoVideo segment is calculated based on the low-level image feature of the video chip key frame.And that introduced above is used as a reference for measuring the relevance of geographic video.(3)This paper proposes a semantic association of geographic video organization model.Based on the semantic segmention of geographic video,the data is divided into three levels: geographical video,video clip and key frame.From the perspective of human semantic understanding,the geographic video is divided into a collection of event elements with complete semantics.The model uses video clips as the basic unit of data organization,according to the video data relevance,video clips are aggragated based on the event cell semantic and achieved the semantic association of the geographical video organization.Finally,this paper experimented with the model and method proposed above.Taking the multi-camera surveillance video of Duke University as the experimental data,the construction of the ontology-based geographic video expression model,target trajectory extraction,geographic video data organization and correlation analysis are validated in model and method.
Keywords/Search Tags:Geovideo, ontology, video data management, correlation analysis
PDF Full Text Request
Related items