Font Size: a A A

Multiple Target Detection And Tracking Based On Intelligent Video Surveillance

Posted on:2014-05-11Degree:MasterType:Thesis
Country:ChinaCandidate:X LiFull Text:PDF
GTID:2268330425456808Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
As people learn more about safety and demand is higher and higher,Intelligent and efficientReal-time video surveillance system is widely favored by the people.Intelligent Real-time videocan analyze the real-time video and make intelligent operation for the abnormal behaviors, suchas the alarm.So the analysis of the moving object in video that is to detect and classify themoving target in image sequence、target tracking、action recognition and the understanding anddescription of the whole scene are important. Detect the moving target in image sequence andtarget tracking are key steps.Due to the independence movement of multiple targets and keep out of each other, makesthe detection and tracking become the difficulties in the study of intelligent video monitoringsystem.For multiple moving targets detection and tracking in intelligent video surveillance, thispaper made the following work:(1) the related knowledge of video object’s pretreatment:image denoising with filters、image edge detection、the processing of he binary image morphology and the simulation analysisof various technique.For view of the median filtering algorithm, this paper proposes a fastmedian filtering algorithm.(2) To test the multiple moving targets:this paper introduced Two-dimensional Otsuthreshold segmentation and solved the error segmentation problem that the target and thebackground difference is small;it Introduced the background difference method and inter-framedifference method in details.For multiple gaussian background model is not strong of real timecapability,this paper put forward a new multiple gaussian background modeling method ofimproved symmetrical frame difference method.(3)Multiple target tracking: This paper introduces the developing situation of the targettracking and some methods simply.then introduce the classical particle filter target trackingprinciple in details.On the basis of the characteristics of color matching,this paper put forward atracking method which based on YCbCr color space the color features of particle filter targettracking method.and then do the Multiple-people real-time tracking in the scene.
Keywords/Search Tags:Intelligent monitoring, Thesy mmetrical frame difference, Multiple gaussianmodel, Characteristics of color matching, The particle filter
PDF Full Text Request
Related items