Point-line RGB-D SLAM Algorithm Based on Multiple Back-projection Strategy

To overcome the limitations of the application of the SLAM algorithm, which is based on point features under harsh conditions, this paper innovatively proposes a SLAM algorithm that is based on point and line feature fusion. The algorithm uses an RGB-D camera as an input sensor to extract feature points and feature lines simultaneously from the input RGB image. To fully exploit spatial line feature information and mitigate the impact of environmental noise, we propose a multiple back-projection strategy guided by a depth map to robustly recover spatial line information. Finally, the proposed algorithm is compared with the ORB-SLAM2 algorithm in the public dataset TUM RGB-D. The results show that the proposed algorithm significantly improves the accuracy of localization and mapping.