Representation, Modeling and Recognition of Outdoor Scenes.
Annual rept. no. 3, Apr 94-Apr 95,
SRI INTERNATIONAL MENLO PARK CA
Pagination or Media Count:
The goal of this project is to advance the state-of-the-art in scene interpretation for autonomous systems that operate in natural terrain. In particular, techniques are being developed for representing knowledge about complex cultural and natural environments so that a computer vision system can successfully plan, navigate, recognize and manipulate objects, and answer questions or make decisions relevent to this knowledge. The results to date include the development of new representations for rapidly modeling terrain from multiple images -- including 3-D hexagonal surface meshes and particle-based 3-D modeling primitives that allow a robotic system to incrementally build a complete geometric model of its environment. We have also extended and improved our set of object recognition primitives to allow reliable labeling of such scene attributes and components as color, texture, shadows, and a variety of linear structures skyline, ridgelines, roads, etc.. The most recent results are detailed in four published papers included as appendices to this report. MM
- Land and Riverine Navigation and Guidance