Research On Lucky Imaging Algorithm In Frequency Domain

Due to the influence of atmospheric turbulence,the wavefront of electromagnetic waves from distant celestial bodies will be distorted,which results in the image quality of ground-based telescopes greatly reduced,and the resolution of imaging is severely limited.In order to minimize the influence of atmospheric turbulence on imaging effect and achieve high resolution imaging of space targets by ground-based telescopes,many effective active and passive techniques have been proposed,one of which is lucky imaging technology.Lucky imaging is a relatively simple after-event image processing technology.This algorithm mainly evaluates the image quality of short exposure images and selects a certain proportion of high-quality images for registration and superposition,so as to obtain high-resolution target images.Nevertheless,some defects exist in the classic lucky imaging algorithm.In this algorithm,images are processed in space domain and selected with the full frame or partial region as a unit without taking into consideration the high-frequency components in some directions,thus leading to insufficient utilization of data information and abandonment of some image details.Based on the traditional lucky imaging algorithm,this dissertation studies the frequency domain lucky imaging algorithm.A feasible and effective grouping scheme for fortunate imaging in frequency domain is proposed.Firstly,the classical theory and technology of lucky imaging in spatial and frequency domain are briefly introduced.Then,the frequency characteristics of short-exposure stellar images on ground-based telescopes are analyzed.Using the idea of lucky imaging and the theory of filter,the image fusion rules of lucky imaging in frequency domain are derived.This improved frequency-domain lucky imaging algorithm is designed and implemented with MATLAB.Firstly,the algorithm carries out Fourier transform on the short exposure image sequence,chooses the image information according to the magnitude of Fourier under each spatial frequency,superimposes and integrates the selected information,and achieves image fusion in frequency domain.Then,the fusion results in frequency domain are inversely transformed by Fourier transform to obtain the reconstructed high-resolution image.Finally,logarithmic transformation is used to enhance the image.In the implementation of the algorithm,we use image array conversion to improve the sorting efficiency,and use image grouping fusion scheme to reduce the computational complexity.In addition,in image preprocessing,an effective dynamic adaptive threshold calculation formula is proposed for detection and elimination of cosmic ray image frames.The feasibility and validity of the algorithm proposed in this paper are verified by a large number of computational experiments on measured astronomical target images.Compared with the classical airspace lucky imaging algorithm,this algorithm has higherinformation utilization rate.The FWHM of the star image in the reconstructed high resolution image is smaller.