%0 Journal Article
%T 视觉SLAM综述
%A 权美香
%A 朴松昊
%A 李国
%J 智能系统学报
%D 2016
%R 10.11992/tis.201607026
%X 视觉SLAM指的是相机作为唯一的外部传感器，在进行自身定位的同时创建环境地图。SLAM创建的地图的好坏对之后自主的定位、路径规划以及壁障的性能起到一个决定性的作用。本文对基于特征的视觉SLAM方法和直接的SLAM方法，视觉SLAM的主要标志性成果，SLAM的主要研究实验室进行了介绍，并介绍了SIFT，SURF，ORB特征的检测与匹配，关键帧选择方法，并对消除累积误差的闭环检测及地图优化的方法进行了总结。最后，对视觉SLAM的主要发展趋势及研究热点进行了讨论，并对单目视觉SLAM，双目视觉SLAM，RGB_D SLAM进行了优缺点分析。</br>Visual SLAM refers to simultaneously localizing itself and reconstructing the environment map using cameras as the only external sensor. The quality of the map created by SLAM plays a decisive role in the performance of the subsequent automatic localization, path planning, and obstacle avoidance. This paper introduced the feature-based visual SLAM and direct visual SLAM methods; the major symbolic achievement of visual SLAM; the main research laboratory of SLAM; and the method of SIFT, SURF, and ORB feature detection and matching, key frame selection. In addition, this paper summarized the loop closure detection and map optimization that removed the accumulated error. In the end, the development tendency and research highlights of SLAM were discussed and the advantages and disadvantages of monocular SLAM, binocular SLAM, and RGB_D SLAM were analyzed
%K 视觉同步定位与创建地图
%K 单目视觉
%K RGB_D SLAM
%K 特征检测与匹配
%K 闭环检测&lt
%K /br&gt
%K visual simultaneous localization and mapping
%K monocular vision
%K RGB_D
%K feature detection and matching
%K loop closure detection
%U http://tis.hrbeu.edu.cn/oa/darticle.aspx?type=view&id=20160606