Building of robust multi-scale representations of LiDAR-based digital terrain model based on scale-space theory

Scale-space theory is a framework originally developed for representing signal and imagery data at different scales through the re-sampling of the original data model. Such a framework allows for explicit, scale-invariant visualization and analysis of this type of data. The implementation of this theory in geospatial engineering is essential especially when processing a high-resolution digital terrain model (DTM) such as that produced from LiDAR data. This would allow for rapid retrieval, analysis, and display of the DTM at application-appropriate resolutions. The DTM is a means of representing the surface of the earth digitally, and it can be a Digital Elevation Model (DEM) or Triangular Irregular Network (TIN) model. DTMs can be produced from a variety of data types including photogrammetric, surveying, Global Positioning System (GPS), and Light Detection And Ranging (LiDAR) data. The availability and popularity of the LiDAR systems, which produce high intensity point clouds require the development of robust techniques for the retrieval, display, and analysis of the data model. This article presents a method based on cubic convolution for robust scale-space representation of DTMs created from LiDAR data.