GNSS-Based SAR Imaging for Range Object Recognizability Improvement Based on Successive Spectrum Extension

Limited recognizability of range objects, caused by low-range resolution, represents a bottleneck for passive Global Navigation Satellite System (GNSS)-based synthetic aperture radar (SAR) imaging. To address this problem, this article proposes an imaging method based on the successive spectrum extension of signals. After initial range compression, autocorrelated signals of a local noise-free replica are generated. Subsequently, both the compressed signal, without the carrier phase, and the autocorrelated local replica are transformed into the frequency domain. Next, to achieve the expected range object recognizability, bandwidth extensions are iteratively performed at both the reference and surveillance channels, using the frequency-domain autoconvolution of the compressed signal without the carrier phase and autocorrelated local replica, respectively. At last, the bandwidth-extended signals in both the reference and surveillance channels are converted back into the time domain, correlation operations are performed between them, and the carrier phase is recovered for generating range-compressed signals with high object recognizability. The effectiveness of the proposed imaging method is evaluated through both simulation and a proof-of-concept field experiment. The outcomes show that compared with the method previously developed by Zheng et al., 2024, the proposed method results in a fivefold improvement in range object recognizability. Moreover, field experiments demonstrate that compared with state-of-the-art range object recognizability enhancement methods, such as the improved Diff2 operator- and incremental Wiener filter-based approaches, the proposed method yields 7.82 and 6.57 dB higher image contrast-to-noise ratios (CNRs), respectively.