A Generalized Lq Regularization With Azimuth-Phase-Compensation Method for mmW Automotive SAR Sparse Imaging

Millimeter-wave automotive synthetic aperture radar (Auto-SAR) holds significant promise for autonomous vehicles systems. Sparse imaging methods have the potential to enhance the quality of Auto-SAR images, such as suppressing sidelobes and noise. However, sparse imaging methods based on convex regularization and the measurement matrix suffer from the bias estimate, huge computational complexity and storage requirements. To address these issues, an efficient mmW automotive SAR sparse imaging via generalized Lq regularization with azimuth phase compensation (Glq-APC) is proposed in this article. First, we propose an APC method to establish an efficient Auto-SAR sparse imaging model. It makes the operation of the measurement matrix approximated only by employing the Fast Fourier Transform operator, which greatly reduces the huge computational complexity and storage requirements of measurement matrix. Then, we propose utilizing the nonconvexity of Glq to minimize the bias effect in this model. Besides, this regularization can also converge to the global optimum, which further improves the reconstruction accuracy. Finally, the methodology is validated on simulations and measured automotive radar experiments. In particular, both numerical and visual results demonstrate that the proposed method yields compelling reconstruction performance while maintaining efficient computing performance.