Ridge Estimation Method For Compressed Sampling Based Super-resolution Hyperspectral Imaging

Nowadays,remote sensing technology has been widely used in military,agricultural,meteorological and other fields.The rapid development of remote sensing technology has benefited from the emergence of spectral imaging technology.Different materials have different spectral characteristics,and people can recognize different materials through spectral images,which is the original purpose of spectral imaging technology.Spectral imaging technology combines the functions of a camera and a spectrometer,and it is widely used for air-to-ground monitoring.The more spectral bands of hyperspectral images,the larger the amount of data to be acquired,and the higher the hardware to be required,therefore the development of hyperspectral images has been limited.The theory of compressed sensing provides a feasible way to break this limitation.According to the theory of compressed sensing,a coded aperture snapshot spectral imaging system(CASSI)was proposed.It improved the traditional slit and spatial scanning based imaging system,which suffers from low luminous flux and thus low PSNR.It has become one of the research hotspots of compressed hyperspectral imaging technology.This paper focuses on the improvement and optimization of the coded aperture snapshot spectral images system.The main research contents include the following aspects:(1)First,we introduce the history of compressional spectroscopy,the basic theory of compressed sampling,and hyperspectral imaging based on compressed sampling.(2)Then,we study the ridge estimation method with low computational complexity and good robustness.We have improved the traditional CASSI system by using low-resolution detectors to obtain aliased measurement data in both spatial and spectral dimensions,and then reconstructing high-resolution 3D maps.We carry out numerical experiments and compare them with the existing methods to illustrate that the method has the characteristics of fast calculation speed and high reconstruction precision.(3)We analyzed the measurement matrix and further discussed the advantages of the ridge estimation method in reconstruction.On this basis,we add the sparse representation method to reduce column correlation of the matrix,and carry out numerical simulation to compare the advantages of the reconstruction algorithm in accuracy and stability.