Analytical Results of Classifying Lidar Data with Topography Preserving Non-linear Autonomous Processing for Bare Earth Extraction

We present an innovative way to autonomously classify LiDAR points into bare earth, building, vegetation, and other categories. One of the most desired commodities for LiDAR collection is a high resolution bare earth product with the same resolution as the input data. The LiteSite algorithm automatically extracts buildings and foliage from an urban scene and generates an accurate bare-earth product. Our inpainting algorithms then fill these voids utilizing Computational Fluid Dynamics (CFD) techniques and Partial Differential Equations (PDE) to create an accurate Digital Terrain Model (DTM). Inpainting allows generation of high resolution bare-earth Digital Elevation Models (DEMs) in high frequency terrain for urban 3-D modeling. Moreover, if buildings in the scene are partially obscured by trees, then the LiteSite algorithm automatically removes these obscurations and inpaints the heights while preserving building edge content where vegetation has been extracted. Inpainting preserves building height contour consistency and edge sharpness of identified inpainted regions. This technology reduces manual editing while being cost effective for large scale global bare earth production. Quantitative analyses are provided using Receiver Operating Characteristics (ROC) curves to show Probability of Detection and False Alarm of ground versus non-ground features. Histograms are shown with sample size metrics. Qualitative results illustrate other benefits such as Terrain Inpainting’s unique ability to minimize or eliminate undesirable terrain data artifacts.