SAR稀疏成像模型参数自适应选择

发布时间：2018-11-08 11:18

【摘要】：合成孔径雷达(SAR)稀疏成像模型中的参数选择对于SAR稀疏成像的性能有重要影响,也是当前SAR稀疏成像研究中的难点问题。已有参数选择方法普遍存在适用于个别模型或者运算量大的缺点。基于最大后验概率估计和贝叶斯推理,提出了一种无需额外先验信息的自适应参数选择方法,所有需要的参数都可从已知的数据中获取。通过推导得到模型参数与信号、噪声方差的关系,避免了对数据进行一系列的训练处理,因此极大地减小了计算量。仿真数据和实测数据处理表明,本文方法在实现了较为精确的参数优化选择的前提下,其计算量远低于贝叶斯信息论准则、L-曲线等已有参数选择方法。
[Abstract]:The selection of parameters in the (SAR) sparse imaging model of synthetic Aperture Radar (SAR) plays an important role in the performance of SAR sparse imaging, and it is also a difficult problem in the research of SAR sparse imaging. The existing parameter selection methods generally have the disadvantage of being suitable for individual models or having a large amount of computation. Based on maximum posteriori probability estimation and Bayesian reasoning, an adaptive parameter selection method without additional prior information is proposed. All required parameters can be obtained from known data. By deducing the relationship between the model parameters and the signal and noise variance, a series of training processing is avoided, so the computation is greatly reduced. The processing of simulation data and measured data shows that the computational complexity of this method is much lower than that of Bayesian information theory criterion and L- curve method on the premise of accurate parameter selection.
【作者单位】：国防科技大学电子信息系统复杂电磁环境效应国家重点实验室;
【基金】：国家自然科学基金(编号:61501473,61490693,61490692)~~
【分类号】：TN957.52

【参考文献】