This proposal describes a method for Radar-based track maintenance of ground vehicles under move-stop-move scenarios. The method hinges on a technique for discovering and modeling radar scattering centers on three-dimensional surfaces (SCAT). Though a sequence of HRR and SAR observations of the target vehicle, the scattering model is generated. We show that this technique can model ground targets in move-stop-move scenarios because these models can successfully predict HRR and SAR returns from vehicle orientations which were previously unobserved. We propose to study the issues involved in the application of the SCAT modeling technique as it applies to both SAR and HRR data. These issues include: SAR image vehicle heading estimation, HRR and SAR fusion, Quick-SAR potentials in data collection, multi-sensor data, and multi-resolution data. The study will also involve the identification of method strengths and weaknesses including sensitivity to pose errors and resolution. Commercial applications include Intelligent Vehicle Highway System, Border Surveillance, Battlefield Surveillance. It is expected that this study will be directly applicable to work to be performed on the AMSTE-II Darpa project. Neural Computing Systems is a sub-contractor to Northrop Grumman for feature aided tracking for AMSTE-II.
