论文部分内容阅读
为了获得理想的跨航向分辨率,现有下视三维合成孔径雷达(DL 3DSAR)成像方法所需天线阵列过长,且阵元数目过多。针对该问题,提出了一种基于Lp正则化的DL 3DSAR成像方法。在分析DL 3DSAR回波信号模型的基础上,构建超完备字典,将跨航向成像过程转化为Lp范数最小化问题,并分析其可行性,最后使用稀疏贝叶斯学习方法对其进行优化求解以获得最终的成像结果。仿真实验结果表明,该方法在保证成像质量的前提下可以将成像所需阵列长度减少为原长度的1/4,或者在相同阵列条件下将跨行向分辨率提高1倍。
In order to obtain the ideal cross-heading resolution, the antenna array required by the existing DL 3DSAR imaging method is too long and the number of array elements is too large. To solve this problem, a DL 3DSAR imaging method based on Lp regularization is proposed. Based on the analysis of the DL 3DSAR echo signal model, an overcomplete dictionary is constructed to convert the cross-heading imaging process to the Lp norm minimization problem and to analyze its feasibility. Finally, a sparse Bayesian learning method is used to optimize the solution To get the final imaging result. Simulation results show that the proposed method can reduce the array length required for imaging to 1/4 of the original length or double the resolution of the interlaced array under the same array conditions while ensuring the imaging quality.