Hinged, Pseudo-Grid Triangulation Method for Long, Near-Linear Cliff Analyses

Abstract

Light detection and ranging (LIDAR) scanners can rapidly collect high-resolution, centimeter-level-accurate point clouds representing topography, suitable for change detection if scans are repeated over time. To perform meaningful volumetric change analyses, point clouds are commonly triangulated to produce continuous, digital terrain models (DTMs). However, DTM creation methods generally require a fixed-look direction tied to a specific plane, which results in less than ideal triangulations when modeling areas with largely varying topography, such as coastal cliffs and beaches. Furthermore, for accurate volumetric change analysis, surfaces must be free of intersecting triangles, have consistent facet-normal orientations, and be free of data gaps (holes). The methodology presented herein produces continuous surfaces without inconsistent normals and minimizes holes and self-intersections. The few intersecting triangles and holes may be quickly repaired using existing algorithms and were shown to be substantially less abundant compared with common surfacing techniques. Finally, the data structuring of this technique significantly shortens processing time, reduces memory requirements, and enables efficient and interactive visualization through both subsampling at varying scales and optimized-view frustum calculations.