Search papers, labs, and topics across Lattice.
This paper introduces HBRB-BoW, a refined hierarchical binary vocabulary training algorithm for ORB-SLAM that addresses precision loss in conventional binary vocabularies. It integrates a global real-valued flow within the hierarchical clustering process, preserving high-fidelity descriptor information until the final binarization at the leaf nodes. Experiments show that HBRB-BoW creates a more discriminative and well-structured vocabulary, improving the representational integrity of visual dictionaries in complex environments.
ORB-SLAM's visual vocabulary just got a serious upgrade: HBRB-BoW preserves descriptor fidelity during training, leading to better loop closing and relocalization.
In visual simultaneous localization and mapping (SLAM), the quality of the visual vocabulary is fundamental to the system's ability to represent environments and recognize locations. While ORB-SLAM is a widely used framework, its binary vocabulary, trained through the k-majority-based bag-of-words (BoW) approach, suffers from inherent precision loss. The inability of conventional binary clustering to represent subtle feature distributions leads to the degradation of visual words, a problem that is compounded as errors accumulate and propagate through the hierarchical tree structure. To address these structural deficiencies, this paper proposes hierarchical binary-to-real-and-back (HBRB)-BoW, a refined hierarchical binary vocabulary training algorithm. By integrating a global real-valued flow within the hierarchical clustering process, our method preserves high-fidelity descriptor information until the final binarization at the leaf nodes. Experimental results demonstrate that the proposed approach yields a more discriminative and well-structured vocabulary than traditional methods, significantly enhancing the representational integrity of the visual dictionary in complex environments. Furthermore, replacing the default ORB-SLAM vocabulary file with our HBRB-BoW file is expected to improve performance in loop closing and relocalization tasks.
