基于达芬奇平台的交通视频检测系统的分析与实现-analysis and implementation of traffic video detection system based on da vinci platform.docxVIP

AbstractWith the rapid growth of number of cars and the increasing pressure of urban traffic,Intelligent Transportation was gradually emerging advantage in many ways to improve transport efficiency.Actually,Intelligent Transportation Systems adjust and control traffic flow reasonably through the extraction of traffic information on road.In recent years,the application of image analysis and computer vision in the field of intelligent video surveillance also get more mature,People at home and abroad also put more and more attention on investment and research in Intelligent Transportation Systems.In this paper,intelligent video analysis methods was used to detect and analyze traffic video on road,to build an embedded traffic video detector based on DaVinci technology.In this paper, DM6446 chip of TIs DaVinci platform was used as core. Firstly, we build an embedded video detection hardware system, analyzing major modules of hardware system, cutting and transplanting the embedded operating system, booting the system. Secondly,In respect of target detection,it study several commonly used detection algorithm in video detection system.Against the characteristics of DaVinci embedded platform,we select continuous three frame difference method to detect target vehicle.In respect of target tracking,the combination method of color histogram and Kalman filter was used to track the target vehicle. Finally, transplant and optimize the algorithm to the DM6446 platform, completed xDAIS algorithm standard package,established the Codec Engine and Codec Server.And gradually optimized the detection algorithm based on the characteristics of DSP platform,to meet the requirements of embedded system.The focus of this paper is how to take use of the DaVinci dual-core resources more effectively,in order to take full advantage of powerful hardware processing and software resources of DaVinci.The experimental results show that this traffic video detector based on DM6446 has good detection result