【摘要】 压缩感知理论为遥感空间超分辨技术提供了一种新的实现方式。其中观测矩阵决定着压缩感知的采样规则，设计和优化观测矩阵对于保证信号的重构品质具有重要意义。文章在并行压缩感知成像系统的基础上，提出了一种基于列独立性和互相关性的观测矩阵优化算法。算法通过正交三角分解增强观测矩阵的列向量独立性，再通过特征值分解和等角紧框架约束来降低观测矩阵与稀疏矩阵之间的互相关性。同时，文章还提出了一种基于阈值分割的优化方法，将观测矩阵转化为便于硬件实现的二值矩阵，在降低了硬件加工难度的同时，进一步增强了观测矩阵的效果，提高了优化算法的实用性。仿真实验证明，文章提出的优化算法相较于传统优化算法在峰值信噪比方面提升1～2 dB，具有更好地优化效果。更多还原

【Abstract】 The compressed sensing theory provides a new implementation method for remote sensing spatial super-resolution technology. The observation matrix determines the sampling rules of compressed sensing, so it is of great significance to design and optimize the observation matrix for ensuring the quality of signal reconstruction. We propose an observation matrix optimization algorithm based on column independence and cross correlation for the parallel compressed sensing imaging system. The algorithm enhances the column vector independence of the observation matrix through QR decomposition, and then reduces cross correlation through eigenvalue decomposition and isometric tight frame constraints. Meanwhile,an optimization method based on threshold segmentation is proposed to transform the observation matrix into a binary matrix that is convenient for hardware implementation. This reduces the difficulty of hardware processing and further enhances the effectiveness of the observation matrix. Simulation experiments have shown that the proposed optimization algorithm improves the peak signal-to-noise ratio(PSNR) by 1-2 dB compared to other algorithms and has better optimization effects..更多还原