Model-Based Autofoceus for Stripmap SAR Images Formed via Convolution Back-Projection

Many autofocus algorithms exist for correcting uncompensated residual phase errors in SAR images. These algorithms depend on the SAR modality (i.e. spotlight, stripmap, etc.). In this paper, we develop a model-based phase error estimation method and apply it to correct the phase error for stripmap SAR images formed via convolution backprojection (CBP). Our phase estimation method uses classical subspace fitting techniques well known in the anay processing literature. The novelty in this paper is how we derive our autofocus method from the stripmap SAR forward model and we show that applying the phase error correction to the very popular CBP algorithm is very natural. We also show that our proposed method is non-iterative in the sense that we do not have to iterate between the image domain and the range compressed domain to obtain the phase error estimates. As our derivation shows, we do not have to form the image to estimate the phase errors