Generating Semantic 3D Models of Underground Infrastructure

摘要

By combining two previously unrelated techniquesȁ4;semantic markup in a scene-graph and generative modelingȁ4;a new framework retains semantic information until late in the rendering pipeline. This is a crucial prerequisite for achieving enhanced visualization effects and interactive behavior that doesn''''t compromise interactive frame rates. The proposed system creates interactive 3D visualizations from 2D geospatial databases in the domain of utility companies'''' underground infrastructure, creating urban models based on the companies'''' real-world data. The system encodes the 3D models in a scene-graph that mixes visual models with semantic markup that interactively filters and styles the models. The actual graphics primitives are generated on the fly by scripts that are attached to the scene-graph nodes.