MMW Dual-Band 3-D Sparse Imaging Using Reweighted $l_{q}$-Norm Minimization With Dimension Breakdown Method

Multiband fusion and spatial sparse sampling are two effective methods to alleviate the high hardware demands for millimeter-wave (MMW) near-field imaging radars. Image degradation incurred has to be corrected by reconstruction algorithms. Several difficulties arise when conventional sparse recovery algorithms are applied, such as the huge sampling matrix. In this work, we propose a 3-D image reconstruction framework for MMW near-field radars with simultaneously band-fusion and spatial sparse-sampling. By introducing a dimension breakdown method and an improved $l_{q}$-norm optimization based on the sparse Bayesian learning, the problem of undersampling in both frequency and spatial domains is decoupled, the huge sampling matrix is avoided, and the compressive sensing problem is solved with less computational time. Simulations and experimental results demonstrate that the proposed method can accomplish MMW dual-band sparse 3-D imaging with high efficiency and high image quality.