结合合成孔径雷达(SAR)图像特点,提出了一种基于压缩感知的SAR图像压缩与重构方法,并给出了具体的方法及 详细流程。该方法首先将原始SAR图像进行分块处理,同时,利用离散小波变换(DWT)对分块结果进行稀疏处理,利用 近似QR分解后的随机高斯矩阵对稀疏处理结果进行低维线性观测,实现了SAR图像的稀疏化表征与压缩。文中讨论的 改进的正交匹配追踪(OMP)算法,与传统的OMP算法相比,改进的OMP算法在保证重构精度的前提下,可有效提高收敛 速度。最后,通过离散小波反变换等处理获得最终的SAR图像重构结果。仿真实验结果证明所提方法的有效性与可行 性。

In this paper, a new SAR imagery compressing and reconstruction method based on Compressed Sensing (CS) is proposed. The detailed step and flow chart of the method are shown based on CS theory and SAR imagery characters. In the method, the SAR imagery can be carved up to several sub-imageries firstly. What's more, Discrete Wavelet Transform (DWT) can be utilized to make SAR imagery sparse. And then the random Gauss matrix after approximate Orthogonal-matrix and Right-matrix (QR) decomposition can be employed to complete the low-dimension measurement and the SAR imagery compressing for sparse results. In this paper, a modified Orthogonal Matching Pursuit (OMP) algorithm is proposed. On condition of the same reconstruction precision, the convergency speed is enhanced by using the proposed modified OMP algorithm compared with the original OMP algorithm. Furthermore, some processing such as inverse DWT and so on can be engaged to achieve the final reconstructed SAR imagery. Simulation results prove the effectiveness and feasibility of proposed SAR imagery compressing and reconstruction method.

国家重点基础研究发展计划(973 计划)项目