This is the first-year funding of a three-year continuing award. Three-dimensional data taken from complex scenes place strong demands on the robustness and generality of visible surface reconstruction techniques. In this research a complete reconstruction algorithm will be developed that distinguishes between surfaces, space curves, and clouds of points based on asymmetries in point distributions in expanding three-dimensional image neighborhoods. The properties of this algorithm will be studied analytically and confirmed experimentally on both synthetic data and range data acquired from a variety of sensors. The software will be made available to other researchers for further testing. The goal is to understand fully both the strengths and weaknesses of the technique. This will provide a solid basis for future research and will make it easy to evaluate the technique's usefulness for specific applications. Potential applications range from robotic navigation to reverse engineering.
