Research And Implementation Of Motion Detection And Tracking In Intelligent Surveillance System

With the rapid development of the computer technology, network technology,multimedia processing technology and electronic technology, intelligentsurveillance system is also getting more and more attention. Compared withtraditional video surveillance, intelligent surveillance system using computervision technology that monitor the scene for target detection and understandingand cause automatic alarm when an abnormal event happened, which greatly savesmanpower, material and financial resources. Especially in recent years, with theappearance of various high-performance processors and intelligent detectionalgorithms, high-performance intelligent surveillance system has also been rapidlydeveloping and is more widely applied in the industrial, transportation, banking,public safety and other aspects.This thesis first researched and improved the moving target detection andtracking algorithm of computer vision, then designed and implemented anintelligent surveillance system which is suitable to the unattended communicationbase station with using Freescale controller I.MX53as hardware platform andusing Linux as software platform. This Intelligent surveillance system integratesvideo monitoring, abnormal event detection, target identification tracking,multi-sensor monitoring, and other functions. The system can meet the monitoringrequirements of communication base stations, public places, power transformersubstations and so on. This main theme of paper is around the monitoring systemof substation system.First, the paper studied the moving target detection algorithm and analyzed itsadvantages and disadvantages and introduced complex Gaussian mixture model.Then proposed an improved Gaussian mixture model based on the complicatedmodeling process and the insufficient real-time character of the algorithm. Thisimprovement used block thought and adaptive Gaussian distribution method.Second, in order to implement the scenario for the monitoring of movingtarget tracking, the paper studied moving target tracking algorithm, elaborated thedetail principle of Kalman filter, and designed multi-target tracking algorithmbased on Kalman filter. Moreover the algorithm can accurately identify andtracking multiple moving targets.Third, the paper designed the platform and device driver of intelligentmonitoring system and customized Linux software platform for intelligentmonitoring system intelligent. Then designed Linux device driver of video acquisition chip TW2835, serial port expansion chip SC16IS752chip and I/Oextension chip PCA9635chip based on monitoring system hardware requirementsThis design is focused on the functional requirements of intelligent monitoring. Allof those completed the construction of the software platform.Finally, to realize the function of intelligent monitoring system, the paperdesigned video acquisition application of intelligent monitoring system on thebasis of the Linux and divided intelligent monitoring system into video processingsystem and engine control system through analyzing the function of intelligentmonitoring system. In this process the system implemented the function of videoacquisition, video storage, abnormal event detection, target tracking, videocompression and sending, etc. At last, the system achieved the feasibility of theabove functions by experimental test.