Research On GPU-based Fourier Transform Spectral Restoration Method

Fourier transform imaging spectrometer is a kind of new detecting instrument which can acquire the image information and spectral information of the target object at the same time.The process of spectral information retrieval from the interferometric image which acquired by the imaging spectrometer is called spectral recovery.Conventional spectral recovery method is difficult to deal with the non-linearity problem in the spectral system because of the non-uniform sampling and the cost of considerable time.In order to achieve the fine recovery of the spectrum and improve the efficiency of spectral recovery to meet the needs of rapid spectral recovery in current scientific research or industrial applications,this paper studies the Fourier transform spectral recovery method based on GPU computing.Firstly,the basic theory of Fourier transform spectrometer is studied.The flow of spectrum recovery is discussed and the system structure and working principle of the spectrometer based on Wollaston prism and retroreflector are introduced.The mathematical model is established to demonstrate that the optical path difference in the system change nonlinearly with the sampling interval.Secondly,the conventional method of spectral recovery based on Fourier transform spectroscopy is studied.The whole process of spectrum recovery--the detrending,apodization,phase correction and non-uniform Fourier transform are described.The basic theory of non-uniform fast Fourier transform(NUFFT)is studied.Simulation spectral recovery experiments based on NUFFT and based on FFT algorithm are carried out.It is proved that the non-uniform fast Fourier transform the transformation can recovery the spectrum more accurately.Thirdly,the non-uniform Fourier transform spectrum recovery method based on GPU is studied.The basic theory of GPU calculation and the basic principle of its acceleration calculation are expounded.The optimization direction of GPU algorithm is put forward.The apodization method based on GPU calculation is studied,and it is verified that the apodization method based on GPU calculation is superior to the traditional apodization method.The basic principle of fast Gaussian uniform grids is discussed,and the performance between the fast Gaussian grid computing based on GPU and the traditional fast Gaussian grid is analyzed.The basic theory of fast Fourier transform(FFT)is studied,and the implementation of fast Fourier transform based on GPU is studied.Simulation results show that the GPU-based FFT algorithm outperforms the traditional FFT algorithm in time performance.An Overlap structure based on the Fourier transform spectral recovery algorithm is proposed,and it is verified that the algorithm using the Overlap structure has more excellent performance than the algorithm without the Overlap structure.Finally,a birefringent polarization interference imaging system based on Wollaston prism and retroreflector is built,and the spectral verification experiment is carried out.The results show that GPU-based Fourier transform spectrum recovery method can quickly and accurately recovery the spectrum of the detection target.This method is suitable for Fourier transform spectral imaging system which requires fast spectrum recovery,small size and low power consumption.