针对目标跟踪过程的双目视觉测距方法研究  

A Binocular Vision Ranging Method Used in Target Tracking Process

作　　者：卢艳军[1] 王诗宇 张太宁[1] 李元龙[1] LU Yanjun;WANG Shiyu;ZHANG Taining;LI Yuanlong(Shenyang Aerospace University,Shenyang 110136 China)

机构地区：[1]沈阳航空航天大学

出　　处：《电光与控制》2020年第1期60-63,101,共5页Electronics Optics & Control

基　　金：辽宁省自然科学联合基金(20170540712)

摘　　要：无人机进行目标跟踪时,获取与跟踪目标之间距离信息的过程,对于算法的实时性有着很高的需求。针对这一需求,提出了基于Matlab工具箱与OpenCV视觉库相结合的双目立体视觉测距方法。该方法包括采用Matlab工具箱对摄像头进行双目标定,以及将标定结果导入到基于OpenCV的立体匹配BM算法并进行后续的测距处理两部分。通过试验验证了该方法具有较好的准确性和优越的实时性,可以满足无人机动态跟踪时的测距需求。When the UAV performs target tracking there is a high demand on the real-time performance of the algorithm in the process of obtaining the distance between the UAV and the target.In consideration of the demand a binocular stereoscopic vision ranging method based on the combination of Matlab toolbox and OpenCV vision library is proposed.The method uses the Matlab toolbox to perform binocular calibration on the camera and then imports the calibration result into the OpenCV-based stereo matching BM algorithm for subse-quent ranging processing.Through experiments it is verified that the proposed method has high accuracy and excellent real-time performance which can meet the ranging requirements of UAV dynamic tracking.

关 键 词：目标跟踪 双目视觉 双目标定 立体匹配 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]