SGOR: Outlier Removal by Leveraging Semantic and Geometric Information for Robust Point Cloud Registration

Guiyu Zhao,Zhentao Guo,Hongbin Ma

2024-07-09

Abstract:In this paper, we introduce a new outlier removal method that fully leverages geometric and semantic information, to achieve robust registration. Current semantic-based registration methods only use semantics for point-to-point or instance semantic correspondence generation, which has two problems. First, these methods are highly dependent on the correctness of semantics. They perform poorly in scenarios with incorrect semantics and sparse semantics. Second, the use of semantics is limited only to the correspondence generation, resulting in bad performance in the weak geometry scene. To solve these problems, on the one hand, we propose secondary ground segmentation and loose semantic consistency based on regional voting. It improves the robustness to semantic correctness by reducing the dependence on single-point semantics. On the other hand, we propose semantic-geometric consistency for outlier removal, which makes full use of semantic information and significantly improves the quality of correspondences. In addition, a two-stage hypothesis verification is proposed, which solves the problem of incorrect transformation selection in the weak geometry scene. In the outdoor dataset, our method demonstrates superior performance, boosting a 22.5 percentage points improvement in registration recall and achieving better robustness under various conditions. Our code is available.

Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

The paper primarily addresses the issue of outlier removal in point cloud registration, particularly in scenarios with simple geometric structures and sparse features, by leveraging geometric and semantic information to achieve robust point cloud registration.

Specifically, existing semantic-based methods perform poorly when dealing with inaccurate or sparse semantic information and only use semantic information for point-to-point or instance-level semantic correspondence generation, which limits their performance in weak geometric scenarios. To address these issues, the paper proposes the following improvements:

1. **Secondary Ground Segmentation and Region-based Voting Loose Semantic Consistency**: Enhances robustness to changes in semantic accuracy by reducing reliance on single-point semantic information.

2. **Semantic-Geometric Consistency**: Fully utilizes semantic information to significantly improve the quality of corresponding points, addressing the issue of heavy reliance on geometric information in previous methods.

3. **Two-Stage Hypothesis Verification**: Solves the problem of incorrect transformation selection in weak geometric scenarios.

Through these methods, the new technique proposed in the paper (SGOR) demonstrates superior performance on outdoor datasets, improving the registration recall rate by 22.5 percentage points and achieving better robustness under various conditions. Additionally, the paper provides a detailed description of the specific implementation steps and technical details of the method, including problem definition, semantic-aware correspondence establishment, outlier removal based on semantic-geometric consistency, two-stage hypothesis verification, and transformation refinement.