Implementation For Vehicle Detection And Tracking System Based On DAVINCI

The intelligent transportation is widely recognized as a trans-century growth point and draws enormous attention from developed countries since mid-20th century. Considering the current status of China’s transportation system, development of intelligent transportation becomes necessary. The algorithm of vehicle detection and tacking have been studied based on OpenCV.so an adaptive mixture Gaussian background modeling method based on improved Kalman filtering theory is proposed. What’s more,exploit traffic surveillance videos to test algorithm, then place the effective algorithm into DVS6446 to finish the research of Embedded traffic detection and tracking system of DAVINC1. The key contents of dissertation pertain to following aspects:1、Initiates an adaptive mixture Gaussian background modeling method based on improved Kalman filtering and use shape and centroid tracking combined with Kalman filter to carry out vedio vehicle tracking. By analyzing and improving traditional mixture Gaussian background modeling, propose an adaptive mixture Gaussian background modeling method. At the same time, compared to the main track algorithms to choose target tracking algorithm which is appropriate for system of in this paper.2、Study the software/hardware structure and development process of DM6446; Construct DM6446 development environment (like Linux development environment, SDK development kit installation and development board startup programs and kernel cutting and burning); complete configuration parameters of development environment.3、The transplant of algorithm and embedded system. The accomplishment of previous target entails these following efforts:study Codec Engine、Codec Server、xDM、xDAIS, etc.; Establish algorithm and engines:produce Codec Server; integrate engines; finally, generate ARM side application.4、Test the system. Collect and test statistics to produce the analysis of false drop and missions; combined with the aforementioned analysis to complete the run chart of traffic flow.