Integrated Color CCD and LADAR Based Object Modeling and Recognition: ACoregistration Approach to Multisensor Target Recognition with Extensions to Exploit Digital Elevation Map Data

摘要

A target recognition capability is described that performs: color targetdetection, target type and pose hypothesis generation, and target type verification by 3-D alignment of target models to range and electro-optical imagery. The term 'coregistration' is introduced to describe target, range and electro-optical (color and IR) sensor alignment correction. Online model feature prediction using 3-D military vehicle models is demonstrated for 3-D vehicle models. All phases of the recognition cycle are shown on near-boresight-aligned electro-optical and range imagery collected at Fort Carson, Colorado. As a step toward integrating constraints from Digital Elevation Maps (DEM), an automated terrain feature prediction and matching capability is demonstrated. This terrain matching is used to refine DEM to ground-looking imagery registration.