An Orthographic Facet Image Library for Supporting Site Model Refinement and Visualization

This paper addresses the issue of combining intensity information from multiple images for analysis of object surface intensities to support model refinement and scene rendering. Given camera and light source parameters for each image, and a 3D CAD model of objects in the scene, the textures of object surfaces are systematically collected into an organized orthographic library. Occlusions and shadows caused by objects in the scene are predicted from the model and associated with each retrieved surface. The problem of combining intensities from multiple images is explored, and the result is an algorithm that produces a unique surface intensity representation that is as complete and consistent as possible. A successful application of this approach to model refinement and scene visualization is shown, using results from the RADIUS aerial image understanding project.