Research And Implementation Of Video-based Vehicle Detection And Tracking Technology

Dynamic target detection and tracking based on video is a hot research topic in the field of computer vision, in aspects such as public security, military guidance, traffic monitoring has been more and more widely used. The effective detection and accurate tracking of the moving vehicle is the core of modern intelligent transportation research part. This thesis, based on summarizing and analyzing the existing vehicle detection and tracking technology, focuses on the movement of vehicles under the fixed camera. Video-based detection and tracking system is easy to install, stable, has a wealth of visual information, represents the development direction of the intelligent traffic monitoring system, is currently a hotspot at home and abroad.In this paper, three key aspects of intelligent transportation system have been researched, they are vehicle detection, vehicle tracking and hardware system implementation, the specific work as follows:1. Vehicle DetectionAiming at the deficiency of traditional dynamic target detection algorithm, a moving vehicle detection algorithm based on three frame difference and edge information has been proposed. This algorithm can extract a complete outline of moving targets, essentially retains all the information of the object, works well to eliminate the vehicle shadows, strong anti-interference, has a better robustness to the changes of environment, improves the efficiency and accuracy of the moving target detection.2. Vehicle TrackingIn view of the traditional Cam Shift algorithm based on the specific color features is vulnerable to the complex environment such as illumination changes, similar background interference, this paper proposes a kind of Cam Shift tracking algorithm based on multi-feature fusion, the local ternary patterns(LTP) texture features and color features are combined to describe the tracked target, effectively improving the robustness and accuracy of the Cam Shift tracking algorithm.3. Hardware ImplementationThe hardware system design and implementation of video-based vehicle detection and tracking technology, relying on the Texas Instruments(TI) company’s DaVinci series TMS320DM6467chip to build the system based on three frame difference combined with edge information and the dynamic target tracking system based on improved Cam Shift algorithm. The system is realized by using testing road illegal situation under different algorithms, test results meet the design requirements, has good practical value.