An Error-Matching Exclusion Method for Accelerating Visual SLAM

TL;DR

This work tackles the time burden of feature matching in Visual SLAM by accelerating GMS-RANSAC through prioritizing high-confidence matches. By computing neighborhood-based credibility via GMS and biasing RANSAC sampling toward these high-confidence pairs, the method reduces the number of required iterations while maintaining accuracy. The approach offers feasibility-based optimization of GMS outputs and reduces RANSAC input samples, achieving an average runtime reduction of $24.13 ext{%}$ across KITTI, TUM desk, and TUM doll datasets with comparable precision and recall to baseline methods. The resulting improvement enhances real-time SLAM performance in diverse environments, making robust outlier rejection more practical for large-scale and high-rate applications.

Abstract

In Visual SLAM, achieving accurate feature matching consumes a significant amount of time, severely impacting the real-time performance of the system. This paper proposes an accelerated method for Visual SLAM by integrating GMS (Grid-based Motion Statistics) with RANSAC (Random Sample Consensus) for the removal of mismatched features. The approach first utilizes the GMS algorithm to estimate the quantity of matched pairs within the neighborhood and ranks the matches based on their confidence. Subsequently, the Random Sample Consensus (RANSAC) algorithm is employed to further eliminate mismatched features. To address the time-consuming issue of randomly selecting all matched pairs, this method transforms it into the problem of prioritizing sample selection from high-confidence matches. This enables the iterative solution of the optimal model. Experimental results demonstrate that the proposed method achieves a comparable accuracy to the original GMS-RANSAC while reducing the average runtime by 24.13% on the KITTI, TUM desk, and TUM doll datasets.

Figures (9)

• Figure 1: Methodological Framework.
• Figure 2: RANSAC schematic diagram.
• Figure 3: Example of brute-force matching result.
• Figure 4: Screening results of different methods after brute-force matching.
• Figure 5: Improved GMS-RANSAC matching results and time under different preset ORB feature point counts.