Multichannel Enhanced Millimeter-Wave SAR Imaging via Low-Rank Tensor-Train Decomposition

Abstract

Millimeter-wave (mmWave) synthetic aperture radar (SAR) has found wide applications in autonomous driving, landslide detection, urban mapping, etc. However, the high propagation loss of mmWave bands and the limitation of transmitting power have led to limited imaging performance for mmWave SAR. In this article, an enhanced SAR imaging framework that combines along-track multiple channels is proposed using a low-rank tensor-train (TT) decomposition method, which is applicable for a co-located multiple input multiple output (MIMO) array or a phased-transmitting-digital-receiving array. First, the multichannel images are stacked into a tensor form after SAR imaging on individual channels and spatial variant array phase correction for each pixel. Then, the low-rank property of tensor stack is exploited and the TT model is utilized to find the redundancy and leverage the intrinsic structure of image stack. In addition, ket augmentation is introduced to exhibit the local data structure more clearly than the original tensor under TT decomposition. Finally, tensor-train nuclear norm is used to relax the NP-hard problem with low-rank constraint and the minimization problem is solved in the framework of alternating direction method of multipliers for enhanced imaging. The proposed algorithm can effectively improve the working distance and image quality of mmWave SAR. Numerical experiments using simulated data of MIMO SAR and measured data collected by a ground-based phased-transmitting-digital-receiving array system are carried out to verify the performance of the proposed algorithm.

Keywords

• Low-rank
• millimeter-wave (mmWave)
• synthetic aperture radar (SAR)
• tensor decomposition