A robust hybrid spectral estimation algorithm for SAR imaging

This paper presents a robust and computationally efficient hybrid spectral estimation algorithm, referred to as Hybrid, for synthetic aperture radar (SAR) target feature extraction and image formation. Hybrid first extracts the target features via a relaxation-based optimization approach based on a flexible data model which uses a complex sinusoid with an arbitrary unknown amplitude and a constant phase in cross-range and with a constant amplitude and phase in range to model each scatterer. Super resolution SAR images are then generated via Hybrid by using the extracted target features and the 1-D APES (Amplitude and Phase EStimation) nonparametric algorithm. The performance of Hybrid is demonstrated with both numerical and experimental examples.