基于压缩感知理论的稀疏遥感成像系统设计  

Design of remote sensing imaging system based on compressive sensing

作　　者：刘吉英[1] 朱炬波[1] 严奉霞[1] 张增辉[1] 

机构地区：[1]国防科学技术大学理学院数学与系统科学系,湖南长沙410073

出　　处：《系统工程与电子技术》2010年第8期1618-1623,共6页Systems Engineering and Electronics

基　　金：国家自然科学基金(6080207960901071);教育部新世纪优秀人才支持计划资助课题

摘　　要：高分辨率的应用需求使得传统的遥感成像系统面临高速率采样、海量数据存储等难以突破的瓶颈问题。基于压缩感知理论设计的雷达和光学稀疏遥感成像系统,突破了Shannon-Nyquist定理的限制,以较少的测量数据实现了同等甚至更高质量的信号重构。首先,根据被测目标和场景的不同特性,分别设计了稀疏表示矩阵;其次,根据互相关最小化原则,选择了与稀疏表示矩阵相适应的最优感知矩阵;最后,研究了适用于二维成像大规模数据的稀疏重构算法。专业电磁散射仿真软件生成的雷达观测数据和复杂场景光学图像的数值仿真,验证了本文设计的稀疏遥感成像系统原理上的可行性。The conventional remote sensing system is faced with some intractable problems,such as high speed sampling and mass data storage,owing to the requirement of high resolution.The synthetic aperture radar and the optical sparse remote sensing systems are designed based on compressive sensing,they break through the limitation of Shannon-Nyquist theorem and realize a equivalent or even better signal recovery based on much fewer measurements.Firstly,the sparse representation matrix is designed according to different characteristics of the measured targets and scenes.Secondly,by minimizing the cross-correlation,the sensing matrix is selected which corresponds to the sparse representation matrix.Finally,a recovery algorithm suitable to large-scale problems is investigated.The feasibility of the designed sparse remote sensing systems is validated by the numerical experiments based on radar echo generated by a professional electromagnetic scattering software and the optical image of complex scenes.

关 键 词：压缩感知 合成孔径雷达 光学成像 稀疏表示 感知矩阵 

分 类 号：TP770[自动化与计算机技术—检测技术与自动化装置]