论文部分内容阅读
Vehicle speed measurement based on traffic video is a widely used method to analyze accidents.In order to solve the problem that the existing algorithms ignore the inter-frame acceleration,a moving-scale vehicle speed measurement was established which considers the vehicle acceleration between two adjacent frames.50 groups of experiments were carried out to verify the accuracy of the algorithm including the videos that the vehicle at different speeds and accelerations.Four groups of experiments were chosen to illuminate the accuracy of the result.The results show that the algorithm can reduce the average relative error 0.2%~0.9%when the vehicle is braking and initial speed is from 40km/h to 80km/h,and when the vehicle is moving under a uniform speed from 40km/h to 80km/h the relative error can be reduced from 0.1%to 0.3%,which means that when vehicles brake at a high initial speed the model is necessary to be used in order to improve the accuracy of results.