HBST: A Hamming Distance embedding Binary Search Tree for Visual Place Recognition

TL;DR

This paper introduces HBST, a binary search tree structure that enables fast, logarithmic-time matching and retrieval of binary descriptors, significantly improving visual place recognition efficiency in SLAM systems.

Contribution

We propose HBST, a novel binary search tree for binary descriptors that achieves efficient logarithmic search and insertion, supported by theoretical analysis and extensive experiments.

Findings

HBST outperforms several state-of-the-art methods in descriptor matching speed.

HBST maintains high accuracy in visual place recognition tasks.

The approach is validated on multiple public datasets.

Abstract

Reliable and efficient Visual Place Recognition is a major building block of modern SLAM systems. Leveraging on our prior work, in this paper we present a Hamming Distance embedding Binary Search Tree (HBST) approach for binary Descriptor Matching and Image Retrieval. HBST allows for descriptor Search and Insertion in logarithmic time by exploiting particular properties of binary Feature descriptors. We support the idea behind our search structure with a thorough analysis on the exploited descriptor properties and their effects on completeness and complexity of search and insertion. To validate our claims we conducted comparative experiments for HBST and several state-of-the-art methods on a broad range of publicly available datasets. HBST is available as a compact open-source C++ header-only library.