基于双目视觉的汽车尺寸检测技术研究

Research on Vehicle Size Detection Technology Based on Binocular Vision

为了快速检测出汽车的尺寸,提高车流量统计的效率,设计了基于MATLAB与OpenCV相结合的双目视觉测量系统。首先使用MATLAB进行双目相机标定,然后将图片导入到OpenCV进行后续的图像处理。应用SGBM算法进行立体匹配,获得被测车辆的视差图并最终得到汽车的尺寸。该系统只需普通的平行双目相机采集左右图像,降低了成本。结果表明,在1 m的物距内,通过SGBM算法求得的尺寸误差可以控制在10 mm之内。

This paper designs a binocular vision measurement system based on the combination of MATLAB and OpenCV.Firstly,the binocular camera is calibrated by MATLAB,and then the image is imported into OpenCV for subsequent image processing.SGBM algorithm is applied for stereo matching to obtain the parallax map of the tested vehicle and finally get the size of the vehicle.The system only needs the ordinary parallel binocular camera to collect the left and right images,which reduces the cost.The results show that,within 1 m object distance,the size error obtained by SGBM algorithm can be controlled within 10 mm.