华中科技大学学报（自然科学版）Issue(z1)：104-108,5.DOI:10.13245/j.hust.15S1026
基于双目视觉的机器人自定位方法研究
Robot autonmouls localization method study based on binocular vision
摘要
Abstract
A robot autonomous localization method based on binocular vision image sequences was studied.Harris corner features were extracted with the binocular vision image sequence and Sobel fil-ter responses of horizontal and vertical was simply compared to each other by using sum of absolute differences error metric,and matching feature points based on the responses value.The improved RANSAC algorithm was used for feature matching optimization.Finally the gauss-newton iterative method was used for solving over determined nonlinear equations of robot pose estimation,and the o-rientation of the robot was realized.Through indoor and outdoor scene experiments,as well as com-paring with the stereo visual odometry algorithm in BoofCV visual library,experiments show that the proposed approach can meet the application requirements in terms of accuracy,stability and run-time.关键词
机器人/双目视觉/特征匹配/位姿估计/定位Key words
robot/binocular vision/feature matching/motion estimation/localization分类
信息技术与安全科学引用本文复制引用
张华,夏菠..基于双目视觉的机器人自定位方法研究[J].华中科技大学学报（自然科学版）,2015,(z1):104-108,5.基金项目
四川省科技支撑计划资助项目（2013GZX0152）． （）
