面向高光谱遥感成像的空谱三维压缩感知方法研究

建立在Shannon采样定理基础之上的现有高光谱遥感成像系统存在信息获取能力和效率不足、数据量大等问题。基于压缩感知的新型成像方法为解决上述问题提供了新思路。但高光谱遥感成像属多维信息获取问题，其更高的数据维数使得相应的稀疏表示问题、投影测量系统及重构问题都变得更加复杂，现有的压缩感知理论不能满足应用需求。本项目针对上述问题，从高光谱遥感图像数据特性出发，研究高光谱图像数据的三维稀疏性度量，在此基础上基于空谱相关性设计结构化投影测量矩阵，并建立空谱稀疏约束下的序列图像重构模型和算法，最后结合可重构条件理论和数值模拟对成像性能进行分析。以此建立面向高光谱遥感成像的空谱压缩感知方法框架，在空谱三维稀疏表示、结构化投影测量矩阵设计、基于空谱稀疏约束的序列图像重构模型等几个方面取得创新。项目成果将为压缩感知理论在高光谱遥感成像中的应用提供科学方法，促进压缩感知多维信息获取理论的发展。

The traditional hyperspectral remote imaging system,which is based on theory of Shannon sampling,is confronted with some problems such as the ability and efficiency issues and huge data. The new imaging method based on compressive sensing provides a new idea for solving those problems. However, the hyperspectral remote imaging belongs to multidimensional information acquisition,the higher dimensions of the data make the sparse representation, measurement system and reconstruction more complicated.The existing compressive sensing theory can not meet the application requirements. In view of the above problems,this project will exploit the characteristic of the hyperspectral image data,study the measure of the 3D sparsity,and then design the structural project measurment matrix based space-spectral correlations,and establish the reconstruction models and algorithms for spectral images sequence and analysis the imaging performance based on reconvery conditon theory and numerical simulation lastly. A well establised new framework for space-spectal 3D compressive sensing for hyperspectral remote sensing imaging, some innovations related to the space-spectra 3D sparsity representation, the design of structural projective measurement matrix and the sequence images reconstruction models based space-spectral sparsity,are expected. The research findings of this project will provide the scientific method for hyperspectral compressive imaging system and promote the development of the compressive sensing multidimensional information aquistion theory.