Tight Integration of Feature-based Relocalization in Monocular Direct Visual Odometry
This work addresses the challenge of achieving globally accurate and smooth camera poses for visual odometry systems, particularly in varying weather conditions, though it appears incremental as it builds upon existing direct methods like DSO.
The paper tackles the problem of improving camera tracking accuracy in monocular direct visual odometry by integrating map-based relocalization, resulting in promising improvements as demonstrated on multi-weather datasets.
In this paper we propose a framework for integrating map-based relocalization into online direct visual odometry. To achieve map-based relocalization for direct methods, we integrate image features into Direct Sparse Odometry (DSO) and rely on feature matching to associate online visual odometry (VO) with a previously built map. The integration of the relocalization poses is threefold. Firstly, they are incorporated as pose priors in the direct image alignment of the front-end tracking. Secondly, they are tightly integrated into the back-end bundle adjustment. Thirdly, an online fusion module is further proposed to combine relative VO poses and global relocalization poses in a pose graph to estimate keyframe-wise smooth and globally accurate poses. We evaluate our method on two multi-weather datasets showing the benefits of integrating different handcrafted and learned features and demonstrating promising improvements on camera tracking accuracy.