Research And Design Of Intelligent Security Video Surveillance System

With social security is becoming increasingly serious, our country need a largenumber of security personnel to do the works of security every year. However, because ofthe attention and energy of people is limited, when abnormal situation occurs, it is prone toomissions, and thus can not react in a timely fashion; the function of the traditional videosurveillance system is single, the video data can only be used as evidence of abnormalbehavior, lack of initiative, and real-time. It can not be together with the work of thesecurity personnel. Therefore, to develop an intelligent security video surveillance systemis of great significance, it has a certain ability to act independently, and be able to assist thework of the security personnel.This thesis analyzes the technical feasibility and market viability of the system. Itfocuses on the intelligent security video surveillance system design, development platformstructures, and algorithm research and system implementation.In the design of this system, video image collection, moving object detection andtracking, face detection and recognition, image compression, intelligent storage and GPRSMMS transceiver modules are combined together. It is a complete set of intelligent securityvideo surveillance system design.In this thesis, the basic principles, the algorithm processes and characteristics aboutfour moving target detection algorithm (Frame difference method, the backgroundsubtraction method based on the median filtering, the background subtraction methodbased on the average and the background subtraction method based on common regional)were studied. In addition, the algorithm about image denoising, the labeling and tracking ofmoving targets were also studied. The principles and characteristics of morphologicalprocessing, the notation of the neighborhood and the Kalman filtering method isexpounded. By experiment, these methods have better real-time and stability, thesealgorithms can be applied to an actual system.In the process of system implementation, we choose the S3C2440microprocessor asthe hardware platform, Linux operating system as the software platform, and QT, OpenCV library is called to realize embedded human-machine interface, image processing, movingtarget detection and image compression.Finally, system design features have been implemented on the Linux platform, verifythe feasibility and effectiveness of the algorithm.