摘要
为改善ORB特征点提取与匹配过程中存在的部分误匹配现象,提高智能移动机器人的位姿估计精度,基于ORB-SLAM2系统框架改进误匹配剔除算法,对RGB图像中ORB特征点的提取进行优化,同时降低关键帧中ORB特征点的误匹配率。实验结果表明,改进误匹配剔除的SLAM算法较RANSAC算法在正确匹配点对数量和运行速度上分别提高了7%、38%。在进行同步定位与地图构建时,采用德国慕尼黑工业大学的公开TUM数据集分别对ORB-SLAM2与改进误匹配剔除的SLAM算法有效性进行验证,后者均方根误差和均值误差较前者分别优化了26%和20%。改进误匹配剔除的SLAM算法不仅能快速处理相机读取的每一帧图像,而且能提高正确匹配点对数量与回环检测精度,从而提升智能移动机器人的位姿估计精度。
In order to improve the partial mismatch phenomenon in the process of ORB feature point extraction and matching,and improve the pose estimation accuracy of intelligent mobile robots,based on the ORB-SLAM2 system framework,the mismatch elimination algorithm is improved,and the extraction of ORB feature points in the RGB image is optimized.,While reducing the mismatch rate of ORB feature points in key frames.The experimental results show that compared with the RANSAC algorithm,the improved mismatch elimination algorithm increases the number of correct matching points and the running speed by 7%and 38%,respectively.When performing simultaneous positioning and map construction,the effectiveness of ORB-SLAM2 and the SLAM algorithm based on improved mismatch elimination were verified through the public TUM data set of the Technical University of Munich,Germany,and the root mean square error and mean error of the latter were optimized by 26%and 20%compared with the former.The SLAM algorithm based on improved mismatch elimination can not only quickly process each frame of image read by the camera,but also improve the number of correct matching points and loop detection accuracy,thereby improving the accuracy of the pose estimation of the intelligent mobile robot.
作者
伞红军
王汪林
陈久朋
王晨
SAN Hong-jun;WANG Wang-lin;CHEN Jiu-peng;WANG Chen(Faculty of Mechanical and Electrical Engineering,Kunming University of Science and Technology,Kunming 650500,China)
出处
《软件导刊》
2021年第8期99-104,共6页
Software Guide
基金
国家重点研发计划项目(2017YFC1702503)。
