虽然采用压缩感知技术(Compressive Sensing, CS)的差分SAR层析成像方法实现了4维空间信息的重构，但是此方法仅利用了目标的稀疏特性并没有考虑目标的结构特性，因此对同时具有稀疏特性和结构特性的目标进行重构时其性能较差。针对这一问题，该文采用联合Khatri-Rao子空间和块压缩感知(Khatri-Rao Subspace and Block Compressive Sensing, KRS-BCS)，提出一种差分SAR层析成像方法。该方法依据目标的结构特性和重构观测矩阵具有的Khatri-Rao积性质，将稀疏结构目标的差分SAR层析成像问题转化为Khatri-Rao子空间下的BCS问题，最后对目标进行块稀疏的l1/l2范数最优化求解。相比CS差分SAR层析成像方法，该方法不仅保持了CS差分SAR层析成像方法的高分辨率特点，而且其重构精度更高性能更优。仿真数据和ENVISAT星载ASAR数据以及地面GPS实测数据的试验结果验证了该方法的有效性。%While the use of differential SAR tomography based on Compressive Sensing (CS) makes it possible to reconstruct the four-dimensional information of an observed scene, the performance of the reconstruction decreases for a sparse and structural observed scene due to ignoring the structural characteristics of the observed scene. To deal with this issue, a method using differential SAR tomography based on Khatri-Rao Subspace and Block Compressive Sensing (KRS-BCS) is proposed. Using the structure information of the observed scene and Khatri-Rao product property of the reconstructed observation matrix, the proposed method changes the reconstruction of the sparse and structural observed scene into a BCS problem under Khatri-Rao Subspace, and then the KRS-BCS problem is efficiently solved with a block sparsel1/l2norm optimization signal model. Compared with existing CS methods, the proposed KRS-BCS methodnot only maintains the high resolution characteristics of CS methods, but also has higher reconstruction accuracy and better performance. Simulations, ENVISAT-ASAR data and ground-based GPS data verify the effectiveness of the proposed method.
展开▼
