3 具备大规模,大尺度,基于直接法的单目SLAM 基于直接法的大范围单目SLAM Large-Scale Direct Monocular SLAM
改动_Labby:大范围——>大尺度
We start by giving an overview of the complete algorithm in Sec. 3.1, and briefly introduce the representation for the global map in Sec. 3.2.
3.1小节,我们首先给出整个LSD-SLAM算法的概述,然后在3.2小节,我们简单介绍LSD-SLAM所用到的地图是如何表示地图的(数学模型)。
The three main components of the algorithm are then described in
Sec. 3.3 (tracking of new frames),
Sec. 3.4 (depth map estimation),
Sec. 3.5 (keyframe-to-keyframe tracking) and
finally Sec. 3.6 (map optimization).
LSD-SLAM算法的三个主要模块,将在如下小节进行详解:
- 3.3小节 新图像的跟踪 (具体来说,就是当有新的图像输入时,它与当前关键帧之间 的变换关系,译者额外添加备注)
- 3.4小节 深度图估计
- 3.5小节 关键帧之间的跟踪,(具体来说,计算关键帧之间的 ,译者额外添加备注)
- 3.6小节 地图优化