Vehicles Distance Measurement Algorithm Research Base On Image Recognition

As people living standard rise, more and more vehicles come into families, but at the same time, some traffic accidents frequently happen, which bring property or life loss to people and nation. In this article, a vehicles distance detection based on the fundamental matrix to lead matching points pairs of vehicle tail features is proposed to avoid the collisions between vehicles based on the existing algorithms. First, the camera calibration is implemented, and then the binocular stereo vision principle is analyzed; Combined with the highway traffic applications, the binocular stereo vision system model is constructed through using the calibration parameters to meet the measurement requirements of the vehicle distance detection.Considering the applications of vehicle distance detection based on binocular stereo vision system, the tail of vehicle is chose as the object to recognize, and then the feature matching is carried based on vehicle tail features, in which the Harris corner detection algorithm is used to detect the vehicle tail corners. Finally, the distance detected is calculated according to the principle of binocular disparity. During the corner feature matching, the OPENCV is used to get initial matching points pairs of the vehicle tail corner features, and then the RANSAC algorithm is performed based on those feature matched points to obtain leading matching points pairs of vehicle tail features.4kinds of distances between vehicles from5meters to15meters are designed during the experiment. Experimental results show that the measurement precision of the algorithm in this paper can meet the need of traffic vehicle safety distance, and it is effective for the vehicle distance detection.