Research On Turbulence Degradation Image Restoration Based On Wavelet Transform

Atmospheric turbulence is an irregular form of motion in the atmosphere.The formation of turbulence is caused by the changes in temperature,pressure and other factors that lead to an irregular change in the density of the atmosphere.That is,the refractive index of the atmosphere is also irregular.Atmospheric turbulence can have a significant impact on optical imaging.Because of turbulence interference,when the optical system through the atmosphere of the target imaging,the observed image will appear point intensity diffusion,image blur,image drift and other turbulence effects.In this way,some operations such as target observation,recognition,tracking and so on,which need to obtain information from the image,are affected.Therefore,reconstructing the turbulence-degraded image is a study of practical value and also a challenging challenge.Because the point spread function of turbulence-degraded images is unknown and the variation is also random,making it impossible to establish a uniform mathematical model.Because it is difficult to obtain exact prior knowledge of such problems,blind image restoration is usually used to study.In order to restore the turbulence-degraded image to a high-quality target image,a turbulence-degraded image restoration algorithm is studied in this paper.A turbulence-degraded image restoration algorithm based on wavelet transform is proposed.The algorithm takes advantage of the time-frequency characteristics of the wavelet.Firstly,the wavelet decomposition of turbulence-degraded images can be used to obtain sub-images of different frequency bands under different decomposition scales.Noise is mainly concentrated in the high-frequency part of the image,so the RL blind restoration algorithm is used to de-blur the low-frequency part,and the high-frequency part is de-noised by the wavelet threshold method.Based on the wavelet coefficients of the high-frequency sub-bands in different directions,the noise variance of each high-frequency sub-band is estimated,and then the adaptive threshold suitable for each band is obtained.Based on these thresholds,The coefficients are contracted,and finally the sub-images of each part are reconstructed by wavelet to realize the restoration of turbulence-degraded images.This not only reduces the difficulty of algorithm calculation and can improve the anti-noise ability of the algorithm.The main tasks completed include:1.Briefly introduced the principle of atmospheric turbulence generation,the influence and significance of turbulence on the imaging system,and the research progress of the problem at home and abroad.2.An overview of image restoration technology.The principles of image restoration and the establishment of degradation models are described.Some traditional image restoration methods and several algorithms of blind restoration are also introduced,and their advantages and disadvantages are analyzed.3.Improved wavelet threshold method.Firstly,the basic theory of wavelet transform is introduced to lay the foundation for the introduction of two-dimensional wavelet transform.The application of wavelet transform in image processing is also studied.It is found that the time-frequency local characteristics of the wavelet transform can flexibly process the details of the image.According to this characteristic,the wavelet threshold denoising method is also studied.However,some commonly used thresholding methods adopt global thresholds and cannot fully utilize the characteristics of the wavelet.Therefore,the self-adaptive thresholding method of Nonnaml Shrink is studied,and an adaptive threshold denoising method applied to the high frequency band of the image is proposed.Through simulation experiments,it is proved that this method has the advantages of keeping image details and edge features more than traditional denoising methods.4.The Richardson-Lucy restoration algorithm in the wavelet domain.The derivation process of Richardson-Lucy algorithm is introduced and the final iterative formula is given.However,due to the influence of turbulence,the point spread function can not be known.Therefore,Biggs' method of estimating the point spread function is studied,and the initial point spread function is calculated using the observed degraded image.Then the initial point spread function and the degraded image are used to find the estimated value of the original sharp image.Then the image estimation at the current iteration time is used to obtain the value of the next moment through this alternate iterative process,to get the best approximation.The original clear image.Finally,the reconstructed restored image is obtained by inverse wavelet transform.And a large number of experiments have proved that the algorithm has a better recovery effect on turbulence degraded images,compared to simply using RL algorithm has a certain improvement in the recovery effect,and also has a better anti-noise ability and faster computing speed.