Research And Application Of Intelligent Vehicle Tracking Method Based On Machine Vision

With the rapid development of intelligent vehicles,target tracking has become one of the main research points of intelligent vehicles.And with the decrease of machine vision's cost,target tracking based on machine vision has been widely used.In this paper,an intelligent vehicle tracking method based on machine vision is proposed,which is used to detect and track vehicles in real time traffic condition.The paper is based on the science and technology project of Shanxi province(20130321005-04).In the process of vehicle tracking,the processed object of the program is every frame in the video,which is an image.We start with the image preprocessing,the aim of which is to choose the pretreating method suitable for operating environment.Then we choose the weighted average method for the image gray processing about the image,and filter the noise in the original image by the median filtering.Finally,we complete the image edge detection using Sobel edge detection operator.The method provides a simple but effective image information for later target detection.To ensure the accuracy of vehicle's road information shooting by the camera,this paper calibrate on the camera using Zhang Zhengyou calibration method.By using the self-made8*10 chessboard calibration board,we obtain the camera's internal parameters,and correct the camera's distortion.Due to the complex environment the vehicle facing,the accuracy of traditional Camshift tracking method is low.The tracking method is invalid when the color of the tracked vehicle is similar to the color of background.To solve the problem,this paper puts forward cross ratio invariant Camshift tracking method,which can detect the characteristic corners about the selected target area by using sub-pixel corner detection method.Then we use the positionalrelation between the characteristic corners and the principle of cross ratio invariance to restrain the selected target.The proposed algorithm is verified through theoretical and experimental method.The feasibility and priority of the proposed algorithm is tested using standard video database.The applicability of the algorithm is proved by the real-time traffic test.The compiling environment is VS2013 and the compiler is based on OpenCV video processing library.We conduct the tracking experiment of vehicle information in standard video database and realistic environment respectively.The experimental results show that the tracking method proposed in this paper is timely and feasible and has higher efficiency.