摘要
针对步进频率连续波探地雷达(SFCW-GPR)目标成像空间的联合稀疏性,在基于稀疏贝叶斯压缩感知(BCS)成像的基础上,提出了多任务贝叶斯压缩感知(MT-BCS)算法;针对不同的任务,采用一种通用的先验层次贝叶斯模型,该算法能在有限观测数据条件下,通过利用每组观测数据之间的相关性,使得从较少的随机样本中恢复原始信号;该算法对每组任务的重构都是独立的,在充分利用观测数据的相关性的同时,又可以保留各自数据的特性,实现各组观测数据之间信息共享;仿真结果表明,MT-BCS的重构性能良好,在相同条件下MT-BCS算法的重构效果优于BCS算法所得到的重构效果。
Based on sparse Bayesian compressed sensing(BCS)imaging,a multi-task Bayesian compressed sensing(MT-BCS)algorithm was proposed for the joint sparse of the target imaging space of stepped frequency continuous wave ground-penetrating radar(SFCW-GPR).For different tasks,a general priori hierarchical Bayesian model is adopted.MT-BCS algorithm can recover the original signal from fewer random samples by utilizing the correlation between each set of observation data under the condition of limited observation data.The algorithm is independent in the reconstruction of each group of tasks.While making full use of the correlation of observation data,it can retain the characteristics of each group of data and realize the information sharing between each group of observation data.The simulation results show that the reconstruction performance of MT-BCS is good,and the reconstruction effect of MT-BCS algorithm is better than that of BCS algorithm under the same condition.
作者
吴冬晖
WU Dong-hui(School of Electrical and Information Engineering,Anhui University of Science and Technology,Anhui Huainan 232001,China)
出处
《重庆工商大学学报（自然科学版）》
2021年第1期57-61,共5页
Journal of Chongqing Technology and Business University:Natural Science Edition
