Integrating topographic features into point cloud processing for terrain modeling

Terrain plays a dominant role in shaping surface processes. Digital Elevation Models (DEMs) are widely used for representing terrain surfaces. With advancements in UAV photogrammetry, high-resolution point clouds are widely used for terrain modeling. However, challenges arise in flat areas due to densely distributed point clouds, and in complex terrains where occlusions and filtering create gaps, affecting terrain modeling accuracy. Considering topographic features are critical for terrain modeling, this study integrates topographic feature elements and topographic derivatives into point cloud simplification and gap restoration for efficient and accurate terrain modeling. This study first extracts topographic features directly from points clouds. Second, the extracted topographic derivatives are used to simplify point clouds, i.e., extracting global control points and local terrain feature points. Third, the extracted topographic feature lines are incorporated into patch matching methods for point cloud gap restoration. The topographic feature-embedded patch matching method enhances its terrain similarity mining capability for more accurate restoration. The experiments results demonstrate that the topographic feature-embedded point cloud simplification and restoration methods improve the accuracy of terrain modeling. This work endeavors to incorporate topographic features into point cloud processing for terrain modeling and can benefit future research related to terrain modelling.