.png){kind=icon}
.png){kind=icon}
Return
A Deep Unfolding Network for Sparse SAR Imaging From Uniformly Downsampled Range-Domain Data
.png){kind=icon}
PRE
AI
DOI：10.1109/tgrs.2026.3714782.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Compressive sensing (CS)- and deep learning (DL)-based sparse synthetic aperture radar (SAR) imaging methods significantly reduce data transmission and storage requirements. However, as required by CS, these methods generally employ the random downsampling strategy. Thus, the radar system must be equipped with high-speed analog-to-digital converters (ADCs) capable of sampling at rates required by the Nyquist sampling theorem. To directly decrease the sampling rate (SR) and the amount of data at the receiver, this article proposes a deep unfolding network (DUN) for sparse SAR imaging from uniformly downsampled data in the range domain. Specifically, an AXB-based linear matrix model for stripmap SAR is first presented to fit the image degradation process from fully sampled data to uniformly downsampled data, which mainly induces global aliasing and amplitude attenuation. By reversing this model, we propose a linear residual matrix (LRM) regularization to suppress aliasing and recover amplitude. Based on the inverse imaging operator, a sparse SAR imaging optimization problem is designed for uniform downsampling with the LRM regularization and the convolutional neural network (CNN)-based regularization. These two constrain the reconstruction of global and local features, respectively. Finally, we unroll the optimization into a two-stage DUN that follows a sequential global-to-local recovery scheme. Experiments on simulated and real stripmap SAR raw data, in both qualitative and quantitative terms, verify the effectiveness of the proposed network for sparse SAR imaging with uniform downsampling in the range dimension for the stripmap mode.
Keywords:
Deep unfolding network (DUN)
linear residual matrix (LRM) regularization
range-domain uniform downsampling
sparse synthetic aperture radar (SAR) imaging
stripmap SAR
Journal
IF:
8.6
Papers:
2.1W
Citations:
10.7W