基于点特征的单目视觉里程计实现  

Monocular Visual Odometry Based on Point Feature

作　　者：夏亮 黄鹤 XIA Liang;HUANG He(School of Geomatics and Urban Spatial Information,Beijing University of Civil Engineering and Architecture,Beijing 102616,China;Beijing Advanced Innovation Center for Future Urban Design,Beijing University of Civil Engineering and Architecture,Beijing 100044,China)

机构地区：[1]北京建筑大学测绘与城市空间信息学院,北京102616 [2]北京建筑大学北京未来城市设计高精尖创新中心,北京100044

出　　处：《北京测绘》2019年第3期304-309,共6页Beijing Surveying and Mapping

摘　　要：本文采用特征点检测实现移动机器人的单目视觉里程计,通过对图像进行网格划分保证特征点均匀分布,提高特征点匹配效率。并行计算基础矩阵及单应矩阵实现单目视觉里程计的自动初始化,采用RANSAC算法剔除外点提高特征点匹配质量。设定运动模型及参考帧模型加速帧间运动估计,并通过合理的关键帧选取策略及图优化提高相机位姿精度及鲁棒性。利用TUM数据集测试本文设计的单目视觉里程计,实验轨迹误差为3.58 m,通过Turtlebot2移动机器人对真实环境进行测试,实验表明本文方法的有效性。The monocular visual odometry of mobile robot is realized by feature point detection.The uniform distribution of feature points is guaranteed by gridding the image,and the efficiency of feature point matching is improved.The basic matrix and homography matrix are computed in parallel to realize the automatic initialization of monocular visual odometer,and the RANSAC algorithm is used to eliminate the outliers to improve the matching quality of feature points.Motion model and reference frame model are set to accelerate the inter-frame motion estimation,and the camera pose accuracy and robustness are improved by reasonable key frame selection strategy and graph optimization.TUM data set is used to test the monocular vision odometer designed in this paper.The error of trajectory is 3.58.Turtlebot 2 mobile robot is used to test the real environment.The experiment results show that the method is effective.

关 键 词：特征点检测 自动初始化 关键帧选取 视觉里程计 

分 类 号：P231[天文地球—摄影测量与遥感]