Model-Based Classification for Laser Radar

"Laser Radar (ladar) sensors are under consideration for BMD ground and interceptor platforms. An advantage of ladar over other modalities, such as SAR, is that it has very high range and Doppler (range rate) resolution. Therefore, ladar has the potentialto separate, characterize, and classify closely spaced objects, which is a critical part of the BMD problem. To realize this potential for complex threat clouds requires advanced scene understanding methods. We propose to apply the model-based approach,where features extracted from the collected data are matched with those predicted from a physical model of the scene. In this SBIR, we develop and evaluate a model-based approach suitable for ladar data analysis that results in a probabilistic physicaldescription of closely-spaced objects in the scene. For BMD, the model parameters of interest include the number of objects, and object shapes, locations, and kinematics. This work will combine the expertise of two companies: ALPHATECH, Inc. willdevelop the model-based scene understanding system, based on our experience as lead algorithm developers in the MSTAR SAR ATR program. The sub-contractor, SPARTA, Inc. provides expertise on ladar phenomenology and in simulating realistic ladar signaturesfor feature prediction, as well as for test and evaluation. The technology developed under this program will contribute directly to the BMDO objective of improving threat object classification