An Efficient Object-Oriented System for Extracting Terrain Features from Contour Maps

Period of Performance: 12/18/1998 - 06/18/1999

$69.7K

Phase 1 SBIR

Recipient Firm

Vexcel Corp.
1690 38th Street
Boulder, CO 80301
Principal Investigator

Research Topics

Abstract

Vexcel Corporation is proposing to build a complete object-oriented system which will interpret a contour map, in order to automatically extract features of interest in military tactical planning. One innovation in our approach to this problem over other highly successful approaches lies in our plan to extend the use of the contour tree approach to aid feature-identification tasks as well as point-based searches; we believe that a heuristic can be designed which relies heavily on the structure of the tree itself as a launch-point for feature identification algorithms. Other innovations lie in the addition of convex-hull algorithms to determine where concavities lie in contours (these are key for identifying important features such as canyons), and in our proposal of methods to handle serious interpretation problems which can arise as artifacts of the map boundary (such as mistaking canyons for depressions). Other techniques, such as contour decimation, are proposed in order to speed up more time-consuming processing tasks. Vexcel's unique contribution to this effort lies also in our understanding of interferometric synthetic aperture radar (IFSAR) data, and in our experience with building commercially viable systems for processing and extracting map features from remotely sensed data. BENEFITS: Vexcel anticipates that the contour map reasoning module will provide an important add-on to at least two of our commercial image processing packages, enabling us to provide its users who also wish to utilize contour map data with the capability to utilize contour maps as well as remotely sensed data.