Vision-Assisted Controlled Imagery Acquisition (VACIA) System

Period of Performance: 09/22/1998 - 09/02/1999

$750K

Phase 2 SBIR

Recipient Firm

Geometrix, Inc.
124 Race Street
San Jose, CA 95126
Principal Investigator

Abstract

Acquisition of metrically-accurate imagery under field condition is of growing importance in DoD application such as 3D scene reconstruction, but current methods are ad-hoc and costly, since no existing products combine metrically accurate imagers with high-precision position/pose data and standardized output data formats. General Reality Corporation (GRC) and subcontractor Korbin Systems Incorporated (KSI) propose to solve this problem by developing a modular, scalable system applicable to a wide range of controlled image acquisition tasks. The proposed system will incorporate a novel vision-assisted approach for obtaining highly accurate geospatial and pose data even during poor GPS reception in urban canyons, by establishing feature correspondences between image frames, and using correspondence information in a Kalman filter to correct inertial measurement device drift and bridge across sparse GPS data. Feasibility was demonstrated during Phase I using a combination of simulation and field testing. During the first year of the effort, GRC and KSI will fully develop the vision assisted navigation algorithms as well as a low-cost inertial/GPS instrumentation board compatible with a wide range of acquisition platforms ranging from low-cost camcorder to high-performance computer based acquisition hosts. An application software suite incorporating the algorithms to output NIMA-standard metadata formats will also be developed. During the second year of the effort, these components will be integrated within a high-performance acquisition system meeting the latest NIMA Video Imagery Standards Profile, and validated against ground truth.