Research On Spatial Heterodyne Interfergram And Spectral De-noising Method

The spatial heterodyne spectroscopy is a new type of spatial modulation ultra-highresolution spectral detection technology developed by Michelson interferometer.With the advantages of ultra-high resolution,real-time performance and high throughput.Chemical evolution and other fields have a wide range of applications and development prospects.The important link of modern data research is mainly the data acquisition process and its denoising processing.Based on the research of the mechanism of the spatial heterodyne spectrometer and the characteristics of the acquired data,this paper analyzes the data identification and de-noising methods.The research objects are laser,Ne lamp,Xe lamp,integrating sphere and water vapor measurement data,respectively.Firstly,based on the traditional de-noising method,analyze the processing flow of the method,start with the inversion spectrum of the spatial heterodyne interferogram,introduce the concept of wavelet transform and its related algorithms and principles,and process the spatial heterodyne spectral data for the wavelet transform algorithm.Be theoretically prepared.In the analysis of traditional processing methods,it is found that there are problems such as improper noise processing and signal distortion.Combined with the advantages of wavelet analysis in extracting signals,classic soft and hard threshold functions are introduced in this method to deal with noise in spatial heterodyne data.In the actual processing process,it was found that the soft and hard thresholding methods of wavelet transform have problems such as excessive smoothing,signal oscillation and noise,so an algorithm based on lifting wavelet transform is proposed on this basis.The algorithm can distinguish signal and noise more accurately in signal decomposition.In addition,threshold selection,two-factor threshold function and median filtering method are introduced in lifting wavelet.During the processing of measured data,it was found that lifting wavelet-based algorithms can remove noise,retain important details and reduce half-wave width.Traditional noise reduction algorithms,wavelet algorithms,and lifting wavelet transformbased algorithms are mostly used to eliminate high-frequency random noise of the data.When in-depth study of spatial heterodyne data,it was found that high-frequency random noise and low-frequency baselines mainly exist in the interference map.Need to deal with low frequency baselines.Through analysis based on two-dimensional Fourier transform,it is found that this algorithm can not only filter out high-frequency noise,but also suppress low-frequency baselines.The algorithm is applied to water vapor data with both high frequency noise and low frequency baseline,and the results show that the effect of removing both noises can be achieved.Then,the in-depth analysis of wavelet and lifting wavelet processing effects is found,and the noise reduction results of the two algorithms will be affected in many ways.Through repeated testing during the noise reduction process,it is found that too many wavelet decomposition layers will filter out useful signals,while too few decomposition layers will cause incomplete noise filtering.From the analysis of the properties of the wavelet basis function,it can be known that various properties will restrict each other,and the parameters of the threshold function can be obtained by adjusting the simulation experiments.Finally,the comparison and quantitative analysis of the noise reduction effects of the four methods show that the two-dimensional Fourier transform algorithm can achieve the purpose of processing high-frequency random noise and low-frequency baseline.It is proved that the two-dimensional Fourier transform algorithm is more feasible in de-noising of spatial heterodyne data.