Research On Fourier Transform Hyperspectral Imaging Restoration Method

Spectral imaging technology,the combination of spectroscopy and imaging technology,acquires two-dimensional spatial images and one-dimensional spectral information by measuring the three-dimensional data cube of the detection target.Owing to high optical throughput and high resolution,Fourier transform imaging spectrometer has been widely used in the fields of environmental monitoring,medical imaging,geological exploration,and military reconnaissance.Interference imaging spectrometers based on birefringent crystals get thousands of interference images through external push sweep.To obtain interference information,it is essential.to quest for a high-precision fast image registration algorithm.in addition,in order to ensure the accuracy of spectral restoration,post-processing of interference data becomes the key.This paper is based on the Savart prism interference spectroscopic imaging system.The interference data processing algorithm of Fourier transform spectroscopy is studied.First,through the foundation of the Fourier transform spectroscopy technology,the working principle,optical path difference analysis and solution method of birefringence polarization interference imaging system based on Savart prism are studied.Secondly,a local up-sampling phase correlation registration algorithm based on sharpening preprocessing is studied.By discussing the theoretical basis of edge extraction algorithm,image sub-pixel registration algorithm and local matrix multiplication,the main flow of fast high-precision registration algorithm is proposed,and the algorithm performance and operation efficiency are analyzed through experimental simulation.Then,according to the characteristics of Fourier interference signals,the detrending terms,apodization and phase correction algorithms in the spectral restoration algorithm are studied.This paper discusses the least square method and the wavelet transform method to remove the trend term,analyzes The spectral characteristics of the apodized functions of the triangular window,the Hamming window,and the Gaussian window,and investigates three phase correction algorithms: modular square root method,Mertz product method and Foeman convolution method.Correct the algorithm and select the best method through experimental simulation analysis.Aiming at the non-linear sampling problem of optical path difference,a non-uniform Fourier transform algorithm(NUFFT)based on fast Gaussian grid method is studied.The feasibility of the method in this paper is verified through experiments comparing the accuracy and speed of FFT algorithm,interferogram interpolation sampling method and NUFFT algorithm to restore the spectrum.Finally,by setting up an experimental device,the spectral calibration and resolution experiments of the system are carried out.In order to verify the performance of the spectral restoration algorithm,it is necessary to perform spectral restoration analysis on the collected external field interference data.