Application And Research Of The Moving Object Detection Technology In The Video Surveillance System

There exists a large number of software systems for video surveillance, these systemsplay an important role in our lives. However, many of the monitoring systems are notrealized intelligent monitoring, monitoring person has been staring at the monitor screenin the monitoring process. These non-intelligent monitoring systems are not onlyconsuming a lot of time and effort,damaging the health of the human visual, but alsothey can not be timely to stop and control dangerous behavior and events. Thereforeintelligent video surveillance system is an urgent needed. The intelligent videosurveillance system can do real-time monitoring in the case of non-monitoring person,intelligently analysis of the image data which captured by the camera,and automaticallygenerate alarms based on the results of the analysis.The moving objects detection is important in intelligent video surveillance system,such as Grid Vedio Monitor System and Advanced Driver Assistance System. Ourmoving object detection algorithm is used in the National Grid Intelligent VedioMonitor System.This detection algorithm provides strong support for automaticreal-time monitoring system’s alarm functions. With the increase of people’s needs, theaccuracy of the algorithms are also increasing.In the thesis, by analysising the Grid Vedio Monitor System, a new method whichcombines the Optical Flow Lucas-Kanade method with denoising processing and theSobel operator is proposed. Traditional Sobel operator and LK optical flow have somedefects, this thesis proposes an improved method. With this algorithm,we can not onlyaccurately detect the multi-objects in the frames sequence,but also can detect themoving direction of the objects. The algorithm of the thesis is mainly used in theNational Grid Intelligent Video Monitor System.The system is mainly used forenvironmental monitoring of the grid, it is realized by a graphical interface library ofQT and digital image processing library of OpenCV. In the monitoring process, whenan unknown object appears in the image, the system can uses the algorithm toautomatically detect intrusion objects and generate an alarm. At the end of the article,we extract the video sequence from the National Grid Vedio Monitor System, use this moving object detection algorithm on the sequence of these frames,and analysis of theexperimental phenomena and data. Experimental results indicate that the algorithm isvery accurate and timely.