基于双目立体视觉的目标物测距研究

Research on Target Ranging Based on Binocular Stereo Vision

为了获取目标物体在空间中的三维信息,本文在双目立体测量原理的基础上,采取OpenCV和Matlab相结合的方法,设计了一个基于双目立体视觉的目标物体测距系统。该系统采取张定友棋盘标定法,使用Matlab内置的工具箱完成双目相机标定,采用SGBM(Semi-Global Block Matching)立体匹配算法在VS2017环境下结合Opencv3.4.1库,对左右相机获得的图像进行立体匹配,生成视差图,将二维空间点重投影到三维空间中,实现二维坐标到三维坐标的转换,即可得到物体的三维点坐标。最终实现通过鼠标点选被测物体视差图后输出选定的空间三维坐标。实验结果显示,该系统的测量精度较高,并且在140 cm距离的测量精度最高。

To get the spatial three-dimensional information of the target object in space, this paper takes the approach of stereo vision measuring principle and designs a target object ranging system based on binocular stereo system. This system is accomplished by the combination of Opencv and Matlab. The binocular camera calibration was accomplished by using the built-in toolbox of Matlab based on the Zhang Dingyou chessboard calibration method. By conducting SGBM(Semi-global Block Matching) stereo matching algorithm adopted in VS2017 environment and Opencv3.4.1, images obtained by left and right cameras are matched in stereo, then parallax map is generated, 2 d space points are re-projected into 3 d space, 2 d coordinates are converted into 3 d coordinates, and 3 d point coordinates of objects can be obtained. Finally, select the parallax map of the measured object through the mouse and output the selected three-dimensional coordinates. Experimental results show that the measurement accuracy of the system is high, and the measurement accuracy is the highest at the distance of 140 cm.