Research Of Vehicle Type Recognition Based On Video

The purpose of Intelligent Transportation System (ITS) include that make full use of the road infrastructure resources, improve the interaction of vehicles, roads and people, enhance security, high-efficiency and comfort of ITS. Intelligent Transportation System (ITS) is base of automatic identification system. With the research of vehicle identification system, research and development of the technologies about intelligent transportation will be accelerated greatly.In this paper, we investigate sports car recognition base on video. First of all, the background of the complexity background sequence images can be obtained using Kalman filter. Then, background difference method and then used to achieve detection of the movement of motor vehicles, and determined threshold by Otsu adaptive algorithm to complete of testing image. The context of finite difference method based on the Kalman filter used in this paper has achieved real-time background extraction. The method had removed false detection which was occured because of slow or fast velocity of moving vehicle, and is insensitive to the microscopic changes of background as well as anti-interference capacity.The algorithm about gray image based on the edge of the shadow of the elimination information is used to remove the shadow of vehicles. The algorithm needn't establish mathematical model and has nothing to do with the light irradiation, eliminate of the shadow of the target in all directions.In Vehicle Recognition System based on Support Vector Machine, extracting RST invariant features avoid camera calibration in traditional features and enhance the generality of system. Finished vehicle identification system based on acyclic graph-oriented decision-making support vector machine, the vehicle types of discrimination under the conditions of small samples is achieved that overcome the traditional problem of local extremum of neural network.Video-based vehicle classification system applicable to complex scenes with large area and multi-objective. Traffic parameter which applied to practical implemente automatic recognition of vehicle and have certain practical value.