Research Of Type-Recognition Techniques For Moving-Vehicles

The main purpose of type-recognition system for moving vehicles is to get variant features of vehicle in real-time, such as vehicle-shape, vehicle-logo, vehicle-license-plate and color of vehicle-body. By utilization of optimized recognition algorithms and information fusion techniques, the vehicle's type, and its color and license plate can be automatically recognized. There are mainly two creative works in this paper, one is color recognition of vehicle-body, and the other is the automatic and real-time location of vehicle-logo. As we all know, the color of light source, the reflectance ratio of vehicle-body, the shading of the surroundings and the camera itself can all influence the color of vehicle-body, so the accurate color recognition is somewhat difficult. In this paper, we present a method of color recognition based on color-difference and color-normalization. Based on the comparative research on different color space models and the color-difference formulas of different color spaces, we find out some practical color space and its color difference formulas. Further on, we improved the recognition precision and its robustness by means of color normalization. Compared with vehicle-license-plate, the vehicle-logo has its own speciality. It is small, similar each other, and is influenced by its size and the light, it lies in different backgrounds and when it belongs to different automobile companies, it is different in size and shape too. So its accurate location is difficult. For the sake of such solutions, a fast and robust method of vehicle-logo location is presented in this paper. By full utilize of the location of vehicle-license-plate, and its correlation of vehicle-logo and vehicle-head, and by utilize of axis symmetry of vehicle-head, an experimental rough searching rectangle of vehicle-logo can be located. Further on, by an adaptive morphology filter and an improved template matching method, the vehicle-logo can be accurately located. "Research of Type-Recognition Techniques for Moving-Vehicles"is the core technique of "Accurate Type Recognition Device of Moving-Vehicles with High Speed", which is supported by innovative fund of ministry of science and technology (the fund number is: 03C26225100257). Based on computer vision, pattern recognition and image processing techniques, we studied such topics as video detection method based on video frequency, vehicle-license-plate location and recognition, vehicle-logo location and recognition, color recognition of vehicle-body, etc. Experimental results show that, the technique presented in this paper is reasonable, and have theoretical and practical values.