Rethinking the Phase in Single-Channel SAR Imagery

Focused single-channel synthetic aperture radar (SAR) imagery is a complex-valued. The common practice in the radar community is to entirely discard the phase information and to only use the detected intensity or magnitude image. This practice hinges on the hypotheses of the so-called fully-developed speckle model which is no more applicable to sub-meter high-resolution SAR imagery. Through properly utilizing the phase information content, not only the spatial resolution present in the complex-valued SAR image is retained, but also the accuracy of detection and recognition of targets can be potentially boosted. Two main contributions are presented in this paper. First, a novel method termed backscatter relative phase image (BRPI) is proposed to characterize the phase content in single-channel SAR imagery. Second, the von Mises (vM) distribution is adopted as a new statistical model for modeling the characteristic phase. The applicability of our proposed algorithm is demonstrated on ground-truthed Radarsat-2 civilian vehicle targets. The proposed method is straightforward to use with any complex-valued high-resolution radar data other than SAR.