Visualization Of Surveillance Video Based On Trajectory

Video surveillance is an important part of the security system in contemporarysociety. The visualization of video surveillance has played a critical role in the safetyof all places. The Surveillance video data is huge, so predict of these data is difficult.The analysis has strong real-time features, the staves who watch and analysis thesesurveillance videos tire easily. Research about surveillance video visualization canoffset the weakness of automated analysis, relief the work of manpower. Throughstudy, we found that in the surveillance video, the video trajectory of the objectcontains a number of important information. As a result, the trajectory analysisvisualization is very conducive to the analysis of the surve illance video, especiallycomprehension of the whole situation and abnormal detection of the surveillancevideo.In this thesis, a novel surveillance video visualization program based ontrajectory is proposed. Users can analysis the surveillance video more convenientlyand effectively in the aspect of the whole situation and the abnormal situation of thescene, which can help staves make decisions. The scheme includes data preprocessing,data modeling based on trajectory cluster, cluster updating based on time, Hermitetrajectory smooth jointing and HSV color model visualization. The data preprocessingis the process of scene data and trajectory data. The trajectory cluster consists oftrajectory dissimilarity calculation, disjoint sets and hierarchical cluster. This analysissolves the problems that demand of input train data first and cluster with the wholetrajectory, which involve speed and direction. Besides, it distinguishes the impactextent of the history and the moment. The trajectory visualization includes clusterupdate, Hermite smooth jointing, HSV color model. The research has beenexperimented with the Puntigam dataset which is extracted from a public transporthub in Graz, Austria and the EVACUATION dataset which is extracted from Miami,Florida.Through the experiment, we proved the design is feasible and effective.