Detection and Tracking of Prominent Scatterers in SAR Data

Tracking prominent scatterers provides a mechanism for scene-derived motion compensation of Synthetic Aperture Radar (SAR) data. Such a process is useful in GPS-denied environments where a lack of precise sensor position data makes standard motion compensation difficult. Our approach to sensor positioning estimates range histories of multiple isolated scatterers with high accuracy, then performs a geometric inversion to locate the scatterers in three dimensions and estimate the platform’s motion. For high-accuracy scatterer range tracking, we first detect prominent scatterers using a CFAR criterion automatic algorithm and then track them with a two-input Kalman Filter (KF) operation. These two steps provide accurate range estimates of multiple scatterers over a sequence of SAR pulses. The KF state space is range and range-rate. We derive data inputs to the KF algorithm from multiple SAR pulses, divided into Coherent Processing Intervals (CPI). Within each CPI, individual scatterer peak amplitudes and phases are available to the algorithm. Our approach to scene-derived motion compensation combines the high accuracy range history estimates with a novel three-dimensional geometric inversion. This geometric inversion uses the range histories to estimate both 3D scatterer location and 3D relative motions of the radar. We illustrate our KF-based approach to high-accuracy tracking and demonstrate its application to estimating scene scatterer locations on synthetic and real collected SAR data.