Research Of Moving Object Detection System Based On Mjpg_streamer Video Surveillance

With the continuous progress of society, the development of more convenient and intelligent video surveillance system in many places replaced the traditional monitoring system.These intelligent, high efficiency monitoring system,in many public places have been widely used.Moving target recognition and detection,as an important part of intelligent video surveillance,is the important foundation and key of the research on target recognition and tracking technology. Research on moving object detection algorithm,from a single algorithm to the multiple moving targets algorithm,not only improved the the effect of moving target detection, and it has very great significance for practical application.The main work of this thesis is to study motion target tracking algorithm in the embedded system application, through the study and improvement on Gaussian mixture modeling model and frame difference moving target algorithm, strengthened the moving target algorithm theory research and the practical application of moving target detection algorithm in mjpg_streamer streamer of video server.The main work of this paper is as follows:(1) Proposed moving object detection algorithm which combined theGaussian mixture model with two frame difference algorithm,through adaptive learning update rates improve Gaussian background modeling,and used the improved adaptive Gaussian mixture model and frame difference results carry out "and" operation,then test the effect of motion target algorithm on VS2010 software platform. Compared with the single Gauss mixed background modeling method, the contour extraction of moving objects has obvious effect.(2)Combined three frame difference method with the background difference algorithm,comparing to ordinary three frame difference algorithm,testing this algorithm on the software platform had achieved the accurate detection effect of moving target detection and overcame "ghosting" phenomenon caused by the single moving object detection algorithm.(3)Based on ARM processor platform,proposed the adaptive threshold setting between three frame difference and background difference algorithm,the advantages of this algorithm as follows:through introducing three frame frame difference threshold adaptive scheme, adaptive threshold can automatically update the background, better reflecting the real-time background effect.the system has stronger adaptability in the hardware implementation.In the process of moving target detection algorithm, through the research and improvement of moving target algorithm,combined with openCV image processing functions, achieving code writing on mjpg_streamer video serversoftware framework,and finally completed threshold adaptive three frame difference and background difference algorithm on mjpg_streamer. Through the image processing algorithm to improve the mjpg_streamer intelligent,achieved the goal of moving target detection. The results of the study have a promotion effect on the development of the intelligent target tracking and behavior recognition.