The Design And Optimization Of Code Aperture On Compressive Hyperspectral Imaging

Spectral imaging,first introduced in the seventies and eighties of the last century,led to a major revolution in remote sensing.Spectral imaging technology is the clever fusion of spectrum detection technology and imaging technology,which can realize the functions of both camera and spectrometer at the same time.It has the important features of "map and spectral".This important property makes imaging spectroscopy one of the most important instruments for achieving Space-to-Earth observations.Different from the traditional spectral imagers,the compressed hyperspectral imaging system combines the traditional spectral imaging technology with the theory of compressive sensing.Today,with an increasing amount of data,compressive sensing make it possible that we can collect part of the data to reconstruct the entire data cube,and restore the original image.At the same time,with the combination of coded aperture imaging technology,the spatial resolution is improved by reducing the pixel size of the coding aperture.Because the coding aperture is added,the difficulty of realizing the detector is greatly reduced.This dissertation focuses on the design and improvement of encoding aperture for compressed hyperspectral imaging system.This research contents the following aspects:(1)First,we introduced the development history of spectral imaging technology and its application.And the designed principle and modeling process of compression hyperspectral imaging system are introduced in detail.Three key points of compression sensing are described.At the same time,the key technologies of compression hyperspectral imaging system are described.(2)Then,we introduce three basic points of compressive sensing: sparse representation,measurement matrix and reconstruction algorithm,the design and improvement of compression perception measurement matrix are mainly studied.Meanwhile,we show the comparison of classical measurement matrix construction method and the different performance of these matrices.Two methods to optimize the measurement matrix based on the matrix decomposition theory are developed.One is the optimization method based on approximate QR decomposition,the other is the optimization method based on approximate singular value decomposition.The applicability of the two optimization methods is compared and found In the basic theory of compressive sensing,these two optimization algorithms have obvious improvement on reconstruction accuracy.(3)After introducing the principle of compressive hyperspectral imaging system and the mathematical modeling process,this paper analyzes and compares the different coding apertures in the system and compares the impact of these apertures on the final reconstruction results of the system.Based on this,the optimization method ofcompressive sensing matrix is applied to the compression hyperspectral imaging system.It is concluded that the optimal fitness of the optimization algorithm in different coding apertures in the system is structured random matrix.