MIMO Radar 3-D Imaging Based on Multi-Dimensional Sparse Recovery and Signal Support Prior Information

Three-dimensional (3-D) imaging using multiple-input multiple-output (MIMO) radar is a new technique which may solve the inherent problems in interferometric inverse synthetic aperture radar (InISAR). But the imaging effect using conventional correlation method is usually poor due to the non-orthogonality of transmitting waveforms with the same frequency. Sparse recovery is hopeful to improve the imaging result by exploiting the sparsity of target 3-D image. However, the existing 3-D imaging method based on sparse recovery has to reshape the 3-D MIMO echo into a vector. So, huge memory spaces and excessive amounts of computation time are required. In this paper, a MIMO radar 3-D imaging method based on multi-dimensional sparse recovery is proposed. First, after splitting and recombining the target 3-D image, the MIMO echo is expressed in a form of tensor-matrix product and a tensor model is built which needs less memory space and computation time than the vector model. In the image tensor of recombined target 3-D image, only limited regions contain the target signal. This prior information of signal support is combined with the signal sparsity to improve the sparse recovering effect. Furthermore, a multi-dimensional sparse recovery algorithm exploiting sparsity constraint and support constraint is proposed to reconstruct the 3-D image. Results of simulated MIMO data and real ISAR data show that, compared with the conventional method, the proposed method can not only save the memory and computation but is superior in the imaging quality.