Abstract: Estimating the distance between two vehicles is very important for transport safety. In order to keep the driving safety and avoid the traffic accidents, the visual-based vehicle detection and vehicle distance measurement are studied. In order to improve the detection precision and shorten the testing time, frame difference is combined with adaboost algorithm to detect vehicle. In the part of distance measurement, a method to calculate the distance between vehicles based on the pixels size of the vehicle's plate area in the image is proposed. Firstly, the image of the rear end of the vehicle is processed for edge detections and vehicle license plate location and rectification; A vehicle distance measurement model is established, the idea of which is to find the relationship between camera and image coordinate system, so as to deduce the vehicle distance model proposed in this paper. Finally, the pixel area of the license plate in the image is obtained based on the model and the vehicle distance measurement experiment is completed. The experiment results demonstrate that the relative error of vehicle distance measurement is less than 7% and the precision can meet the application for vehicle's safety assistance driving system.

Key words: Inter-frame difference, Adaboost algorithm, Vehicle detection, Vehicle distance measurement