The Fast Realization Of Parameter Estimation Of Newton's Rings Interference Fringe Pattern Based On FRFT

Optical measurement is a complex precision measurement technique that analyzes and computes a series of static or dynamically continuous fringe patterns.The digital analysis method of the interference fringe patterns includes two classes based on multiple interference fringe patterns and single interference fringe pattern.At present,the phase shift method is the most common interference fringe pattern analysis method because of its high measurement precision,but this method requires phase shifter,which is not conducive to dynamic measurement.The Fourier transform method can overcome this problem because it requires only one interference fringe pattern,but it can only analyze the frequency domain characteristics of the pattern and needs filter and reverse transformation,and the calculation is large.The fractional Fourier transform is a generalization of the Fourier transform,with the properties of time domain and frequency domain.The chirp signal has the characteristics of energy aggregation within the fractional Fourier transform domain with the matching fractional order,therefore,the fractional Fourier transform is suitable for detecting and estimating chirp signals.In the optical interferometry,Newton's rings interference fringe pattern with chirp signal characteristics are very common,the parameter estimation of the Newton's rings interference fringe pattren with fractional Fourier transform has high accuracy,good antiinterference performance and doesn't need the filter and reverse process.After studying the parameters estimation scheme of Newton's rings interference fringe pattern based on fractional Fourier transform,the main contents of this paper are as follows:(1)Analyze the algorithm implementation process,the Open CV and CUDA are applied to the image processing parts in the implementation reasonably.The parameter estimation algorithm of the Newton's rings interference fringe pattern based on fractional Fourier transform mainly includes the reading,preprocessing,matching fractional search and physical parameter estimation process.In this paper,we use Open CV to realize the image reading,gray-scale and normalized part of the algorithm to improve the simplicity,reliability and accuracy of the application,Meanwhile,in order to further improve the program performance,CUDA architecture was introduced into the algorithm implementation,and the GPU high parallel computing capability was fully utilized to improve the algorithm.(2)Propose a method of block linear superposition,according to the linear superposition property of fractional Fourier transform,the size of interference fringe pattern is reduced and the efficiency of the algorithm is improved.Finally,the verification results are given by the actual Newton ring interference fringe pattern.(3)A search method based on down sampling is proposed,downsampling removes the redundancy that reduce the amount of calculation.Use the matching fractional order of the downsampling image instead of matching fractional order of the original image for parameter estimation;then the matching fractional order is modified in order to further improve the precision.Finally,the verification results are given by the actual Newton ring interference fringe pattern.