ORB_SLAM2 论文与代码总结

最新推荐文章于 2023-06-26 13:53:21 发布

CS_Zero

最新推荐文章于 2023-06-26 13:53:21 发布

阅读量443

点赞数

CC 4.0 BY-SA版权

分类专栏： SLAM 计算机视觉算法文章标签： SLAM CV

本文链接：https://blog.youkuaiyun.com/CSUzhangpeike/article/details/80201777

计算机视觉算法同时被 2 个专栏收录

11 篇文章 ¥19.90 ¥99.00

订阅专栏

SLAM

4 篇文章

订阅专栏

本文深入解析ORB_SLAM系统，涵盖其论文关键点、代码实现细节及运行流程。介绍该系统如何通过跟踪、本地映射与闭环检测三个核心线程实现视觉SLAM功能。

ORB_SLAM论文与代码阅读总结

文章目录

ORB_SLAM论文与代码阅读总结

论文链接：ORB_SLAM Paper

论文关键点笔记

1. Main thread is for tracking, the `System` make 3 more threads for local mapping, loop-closing and viewing.
2. In TRACKING, a local visible map is retrieved using the covisibility graph of keyframes that is maintained by the system，local visible map is used for camera pose refinement;  This local map contains the set of keyframes K1, that share map points with the current frame, and a set K2 with neighbors to the keyframes K1 in the covisibility graph. TRACKING decides inserting keyframes, conditions: >20 frames from relocalization or last keyframe; track features > 50, or <90% of points of reference keyframe.
3. New correspondences for un

了解本专栏