Camera Network Moving Vehicle Detection And Tracking Based On Multi-features Fusion

The video surveillance technology has been widely used in intelligent transportation systems. The use of computer vision technology to transport system of information collection has become an essential issue in the field of intelligent video surveillance technology, and the moving vehicles detection, identification and tracking is the most basic parts. Therefore, this article conducts a related research about the vehicles detection and tracking on the camera.Through analyzing and summarizing the development and application prospects in the Intelligent Transportation Systems, This paper proposes a fusion algorithm based on multi-feature, using this algorithm for the moving vehicle matching and recognition and tracking on the camera network. Three features are the Annular color histogram based on HSV color space, NMI feature and SIFT feature. Proposing a method of vehicle detection and tracking on the sight of Single-camera based on Gaussian mixture model and GM (1,1). About the shadow detection, using the algorithm of chrominance distortion based on the YUV color space and order gradient model based on Sobel operator.