Research And Design On The Network Video Monitoring System Based On Embedded Linux

With the development of big data and cloud computing, the Internet increasinglymature and widely spread in all aspects of people’s life. Video surveillance as a meansof support becomes increasingly important for future safe city and smart city.This topic designs a network video monitoring system based on embeddedLinux.The whole system adopts B/S structure and uses samsung S3C2440as coreprocessor of video monitoring. General computer as the monitor display and controlterminal, using the client browser directly access to the server, and finally display inPC monitor screen. Firstly, building embedded Linux system platform in developedplate structures, the transplantation of U-Boot and the Linux kernel and creating theroot file system, and on this basis for the design of network video monitoring system.Network video monitoring system is composed of five modules, video capture module,video compression module, video transmission module, the Web server module andclient modules. Video acquisition module use Linux provides V4L2applicationprogram interface provided by Linux video data collection; Then after using H.264video compression module to compress video data; And then to transmit video datausing streaming media transmission protocol; The client can request a Web server.The video data are transmited to the client, when Web server receives the request.Network monitoring system design is completed, and for three kinds of movingtarget tracking algorithm to do some research. Tracking window size can’t adaptiveadjustment using MeanShift algorithm; CamShift algorithm can automatically adjustthe tracking window size to accurately track the target, but if the target color andbackground color close or moving targets is completely blocked, will cause the failureto track; Improved CamShift target tracking algorithm based on Kalman filtering maybe good to solve the color similarity or moving targets to be completely blockedthrough time to update and status updates to predict the next frame the location of themoving target, and it can accurate tracking target in complex environment.