Near range radar imaging based on block sparsity and cross-correlation fusion algorithm

In this paper, we propose a novel processing model for compressive sensing (CS)-based stepped-frequency continuous-wave (SFCW) radar near range imaging, which takes the azimuth dependence of the reflection coefficients of targets into consideration. Based on the block sparse property of the received signal in the defined dictionary, 2-D images of the targets can be obtained at each spatial sampling point. A cross-correlation method is then employed to fuse these 2-D images to obtain the final result. Random undersamplings of frequencies and spatial sampling points are conducted to reduce the data acquisition time, the data size, and the computational complexity. Experimental results of an SFCW-based MIMO radar and a ground-based SAR system show that, compared with the conventional matched filtering-based methods, the proposed method can provide artifacts-reduced higher resolution images by using reduced frequencies and spatial sampling points. We also demonstrate that, compared to the conventional CS-based methods, due to the more suitable established observation model, the proposed method can achieve better imaging results with fewer artifacts for near range targets.