The Research Of Vehicle Recognition Method Based On Linear Array CCD

Intelligent transportation system(ITS) is an effective way to solve traffic problems, has become the main developing direction of modern highway transportation system, and has achieved a wide range of applications in many countries.The intelligent charging system is an important part of the ITS. It can effectively alleviate the pressure of the toll because of its automatic charging method. The key of intelligent charging system is the vehicle type recognition, and in the different recognition methods, the vehicle classification is different because of different acquisition of the characteristics. This paper proposes a vehicle recognition system based on linear array CCD, and vehicle is divided into large trucks, vans, buses, mini buses, off-road vehicles, three sedan, hatchback sedan and so on. Compared with the area array CCD, linear array CCD has single image background, high dynamic range, and it can be installed relatively near the vehicle. On an other hand, its image has deformation, so it is necessary to calculate the deformation coefficient. The work of this paper includes following aspects:Firstly, calculate the vehicle geometry parameters using texture features, which include length, width, height, wheelbase and so on. Because the background of the image took by linear array CCD is simple, so the profile of the vehicle is extracted from the image through background difference method, thus the vehicle height can be obtained. Because of the imaging difference between tire and its steel ring, the outline of steel ring can be detected, thus the number of axis, wheelbase and length deformation can be determined. In addition, the length of the vehicle can be obtained through distinguishing vehicle lamps. On the basis of length detection, the width can be obtained through license plate recognition.Secondly, combined parameters can be determined based on basic parameters, such as ratio parameters. Because of its characteristics of dimensionless, the camera calibration is so simple. The combined parameters this paper selected include length height ratio, length width ratio, tail height and vehicle height ratio, wheelbase length ratio and number of axles.The last, classifier can be obtained through lots of calculation of sample based on support vector machine(S VM).Through a lot of experimental verification of various weather conditions, the conclusion is that:the system has fast measuring speed, high accurate rate and weak interference of weather. Besides, it is easy to install and maintain.