RKHS-BA: A Semantic Correspondence-Free Multi-View Registration Framework with Global Tracking

• DOI: 10.48550/arxiv.2403.01254
• Publication Date: 2024-03-02
• Authors: Ray Zhang; Jingwei Song; Xiang Gao; Junzhe Wu; Tianyi Liu; Jinyuan Zhang; Ryan M. Eustice; Maani Ghaffari
• Keywords: Tracking (education)

This work reports a novel Bundle Adjustment (BA) formulation using Reproducing Kernel Hilbert Space (RKHS) representation called RKHS-BA. The proposed is correspondence-free, enables the BA to use RGB-D/LiDAR and semantic labels in optimization directly, provides generalization for photometric loss function commonly used direct methods. RKHS-BA can incorporate appearance within continuous spatial-semantic functional that does not require via image pyramids. We demonstrate its applications sliding-window odometry global LiDAR mapping, which show highly robust performance extremely challenging scenes best trade-off of accuracy.