Robust Tunnel Point Cloud Matching: A Coarse-to-Fine Framework Integrating PCA and GICP with DBSCAN Noise Reduction

View at publisher Summarise with VIRAT
Authors: Jianxin Wu, Ruiyang Zhang, Haoliang Yang, Lin Yang, Junwei NIE, Diyang Chen, Qiang Li
Cited by: 0
References: 17

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

This paper proposes a coarse-to-fine matching framework integrating physical priors with probabilistic optimization: First, a physically parameter-driven DBSCAN clustering denoising algorithm is constructed. Based on clearance requirements in tunnel electromechanical installation specifications, the clustering radius is adaptively determined to effectively strip structured noise closely adhering to the wall surface. Second, Principal Component Analysis (PCA) combined with gravity vector constraints and density skewness detection resolves the$\mathbf{1 8 0}$-degree ambiguity in the principal axis direction, enabling fully automated initial pose correction. Finally, the Generalized Iterative Nearest Point (GICP) algorithm is introduced. By constructing a probabilistic model of the local covariance matrix, it transforms the matching problem into a plane-to-plane maximum likelihood estimation problem. This model effectively suppresses axial slippage through a mathematical mechanism utilizing high-weight constraints on normals and low-weight tolerances for tangent planes. Field validation in a large-diameter submarine shield tunnel in Shandong demonstrated that this method significantly reduced the matching root mean square error (RMSE) from the standard ICP of 0.0185 meters to 0.0109 m, while increasing the overlap rate to 84.2%. Comparisons with the Normal Distribution Transformation (NDT) algorithm demonstrate that the proposed method maintains sub-centimeter accuracy while exhibiting superior environmental adaptability, providing robust technical support for the digital operation and maintenance of tunnels throughout their entire lifecycle.

Fields of Study & Subjects

3D Surveying and Cultural Heritage Robotics and Sensor-Based Localization 3D Shape Modeling and Analysis

Publication details

Journal / source
International Conference on Smart City and Information System (ICSCIS)
Pages
60–69
Publication date
30 January 2026
Language
EN