Design And Implementation Based The DM642Video Surveillance People Counting

Because of video surveillance technology can provide more intuitive and richerinformation content which plays and important role in social management and detection ofcases.When every city across the country spend a huge amount of money to build its own videosurveillance network, shouldering the important mission in public areas of the city, securitynormal monitoring management of the important place and major events security work as wellas the command task of real-time monitoring of major emergencies.Although video surveillance provides intuitive, rich, intelligent, timely image content tosupport the public security system which conducts the investigation managementdecision-making work. With the public security information system construction beingcomprehensively promoting, the requirements of video surveillance system also will beimproved, and the intelligence of video surveillance gradually get attention.The new requirements of the public security departments of the video surveillance systemextensions are how to achieve the intelligence of number Statistics in video surveillance, toprovide accurate and efficient assembling data information to public security departments forcommand decisions. In recent years, along with the rapid development of computer and videoimage processing technology which makes the intelligence of people counting system ofvideo surveillance become possible.This article based on the security video monitoring, designed a system which can implement people counting system of video surveillance. This video surveillance is mainlyusing CMOS camera and DM642DSP (Digital Signal Processing) processor core structuresimage acquisition system, video images will be collected in DM642image to conduct animage segmentation and contour extraction in order to obtain accurate contour data of thetarget task and tracking it to get the target position data. Through reasonable algorithmfinishes counting statistics work, meanwhile, the processed image in real-time return to thehost computer to view which it’s easy for human-computer interaction. In the study based onthe process of DM246which is to achieve people counting, this paper focuses on thefollowing points were studied and discussed:(1) According to the design needs of the system hardware selection planning. SelectCMOS high definition camera, then study and research based on the camera and a DM642video port interface, design the RAW image acquisition mode driver, and using the DM642NDK (Network Development Kit) based on the hardware design, reasonably divide systemprogram into several tasks in order to conduct image data processing, data transforming anddata receiving, etc.(2) According to the learning of theoretical knowledge on the number of statisticalalgorithms and the actual situation, choose a reasonable image processing video surveillancealgorithm based on DM642. Using the background subtraction method to perform on theimage before and after view separation and binarization processing to obtain a target edge;after using improved method on shadows detecting and eliminating, obtain initial targetcontour; finally, using an improved active contour model approach-greedy method to getmore accurate target contour.(3) In order to achieve the target of matching and associated between the frames aroundthe video image, use contour-based target tracking to achieve the real-time location data of thetarget after accurately tracking the moving target; through rational design algorithm countingmethod to achieve people counting.Finally, the design of the people counting systems based on the DM642easy to implementand experiment, obtained better effect, and summarized the deficiencies and defects, Lookedto the next step in the research direction.