Point cloud-based morphological analysis and surface reconstruction of extraterrestrial lava tubes

Lava tubes provide vast and structurally stable underground spaces with minimal temperature fluctuations and effective shielding from cosmic radiation, making them promising candidates for extraterrestrial base construction. Assessing their feasibility requires detailed morphological analysis and high-precision three-dimensional surface reconstruction. This paper proposes a point cloud-based surface reconstruction method. Principal Component Analysis (PCA) is used for adaptive normal estimation and normalization, while an anisotropic weighting function combined with the Moving Least Squares (MLS) projection is used to resample curvature-sensitive regions. An implicit surface is then reconstructed using the Poisson algorithm to generate a high-quality triangular mesh. The proposed method is validated using an Indian lava tube dataset. Twenty-seven typical cross-sections are uniformly extracted along the central axis of the lava tube constructed by the uniform drift algorithm, with their morphological characteristics analyzed. After that, ten point cloud regions with typical features of the lava tube were extracted for surface reconstruction and error analysis. The effect of the initial neighborhood radius (30-80 mm) on normal vector estimation was investigated. The results indicate that the geometric similarity error of point cloud resampling can be minimized to 2 mm, which indicating that our method can reconstruct a surface with high accuracy and preserve local structural details. Furthermore, it provides a theoretical basis and methodological support for the site selection of extraterrestrial bases.