稀疏信号处理已经在雷达目标探测领域得到应用，并获得了优于传统方法的探测性能。然而，雷达目标探测过程中往往存在目标运动、雷达轨迹误差等未知因素，这导致预先设计的字典矩阵无法实现雷达信号的最优稀疏表征。该文将介绍字典学习的一个分支—参数化稀疏表征，该方法通过构建参数化的字典矩阵，实现了对雷达探测过程中未知参数的动态学习和雷达信号的最优稀疏表征。该文还将介绍参数化稀疏表征在逆合成孔径雷达成像、合成孔径雷达自聚焦、基于微多普勒的目标识别等若干雷达探测问题中的应用。%Sparse signal processing has been utilized to the area of radar sensing. Due to the presence of unknown factors such as the motion of the targets of interest and the error of the radar trajectory, a predesigned dictionary cannot provide the optimally spare representation of the actual radar signals. This paper will introduce a method called parametric sparse representation, which is a special case of dictionary learning and can dynamically learn the unknown factors during the radar sensing and achieve the optimally sparse representation of radar signals. This paper will also introduce the applications of parametric sparse representation to Inverse Synthetic Aperture Radar imaging (ISAR) imaging, Synthetic Aperture Radar imaging (SAR) autofocusing and target recognition based on micro-Doppler effect.
展开▼
