Smooth, Piecewise-Polynomial Terrain Representation Using Nontraditional Metrics

Current methods for representing terrain data utilize standard metrics (least squares or maximum deviation) to generate an approximation of terrain surfaces. These methods typically represent the data through a piecewise linear (polygonal) surface which fails to capture the smoothness of the data and thereby creates undesirable artifacts. The result is a poor representation of the actual surface while still requiring high processing and large amounts of memory. By incorporating non-traditional metrics matched to the specific environments (e.g., dense urban, suburban, and rural mountainous) and by utilizing smoother surface representations, it is believed that better surface representation requiring less processing and memory will ensue. This technology has broad military application, such as the real-time visualization of remotely sensed data for unmanned vehicles & operational planning, battlefield simulation, and data feeds to the warfighter on the ground. The Schafer / University of South Carolina Team is proposing an initial study to identify the optimal metrics for various environments followed by a software prototype development effort that implements the surface generation techniques for specific military applications. This effort will leverage Schafer's extensive military understanding and software development experience, along with USC's existing terrain generation algorithms to provide a thorough analysis of nontraditional metrics.