Font Size: a A A

Research On Surveillance System Of Moving Object Detection And Tracking

Posted on:2016-07-29Degree:MasterType:Thesis
Country:ChinaCandidate:Z H LiuFull Text:PDF
GTID:2308330461493997Subject:Circuits and Systems
Abstract/Summary:PDF Full Text Request
Video surveillance has been widely used in people’s life. You can see video surveillance devices in subway and buses, on roads, as well as at hospitals, banks, communities and factories. Video surveillance is now regarded as an important barrier to keep people safe. Thus it plays a more and more important role in life.Traditional video surveillance methods require large amount of manual labor, asking for 24-hour uninterrupted watching on the video.And because of the tedious work, the observers may feel tired and difficult to focus on the screen, thus causing false positive or false negative errors.To solve the problem, intelligent video surveillance systems are studied, in which moving objects detection and tracking techniques are the core techniques. Thus an in-depth study of moving objects detection and tracking techniques has practical significance. In this paper, I introduce my work in the following areas :(1) Moving object detection algorithm. A analysis and comparison of Optical flow estimation, frame difference algorithm and background difference algorithm is shown through experiments. And an improved moving object detection algorithm based on the combination of background difference algorithm and frame difference is brought out. The algorithm is verified under the VC 6.0 platform using OpenCV library. It is proved that the algorithm can remove the empty holes in pictures of the objects, and the detected moving objects are more complete.(2) Moving object tracking. Common methods of moving object tracking are compared and analyzed. Specifically, the working principle of Camshift algorithm is illustrated. There is however a defect in the algorithm. That is a manual initialization of the tracking windows isrequired, thus the algorithm is only semi-automatic. Combining with the moving object detection algorithm I mentioned above, an improved Camshift algorithm realizes full-automation of moving object tracking.(3)In this paper, a moving object detection and tracking system is built on a hardware development platform based on the Realtime ICETEK-DM642-BR development board. The moving object detection and tracking method introduced is transplanted to the DM642 board.Experiments show that the system can be used to detect and track single moving object automatically in simple static background. The system works in real time and is quite stable.
Keywords/Search Tags:Object Detection and Tracking, Camshift, OpenCV, DM642
PDF Full Text Request
Related items