Autonomous Multi-Sensor Registration Toolkit

ALPHATECH proposes to develop a modular system for building and evaluating autonomous multi-sensor geo-registration technologies based on a rigorous mathematical framework and the utilization of a multi-attribute geospatial database. The multi-sensor geo-registration problem is an estimation problem where the objective is to estimate the sensor collection (i.e., sensor/platform position and attitude) parameters that result in the best spatial correspondence of information between the collected sensor data and a model of the real world as represented by a geo-referenced 3-D multi-source fiducial dataset. In Phase I, ALPHATECH developed a multi-sensor, multi-feature representation for the reference dataset and demonstrated key component match algorithms, sensor model updating, and end-to-end registration results. In Phase II we will implement a fiducial database and algorithms to: 1) build experimental fiducial databases over representative sites focusing on FLIR, SAR and GMTI data; 2) develop an experimental infrastructure to automate match (registration) performance evaluation; 3) develop additional match algorithms; 4) develop pre-screening algorithms to determine scene content to adapt registration strategy; 5) perform extensive registration performance evaluation using government supplied multi-sensor datasets; and 6) perform timing and software complexity analysis and algorithm refinement to address issues in eventual integration of this technology in fieldable systems (onboard and ground-based).