Research On Restoration Algorithm For Atmospheric Turbulence Degradation

With the development of spatial target identification,deep space surveying and mapping and earth observation of high-resolution remote sensing,the disturbance effect of atmospheric turbulence has become one of the significant factors which restricts the imaging quality of optical system.During the process of light passing through the atmospheric turbulent medium,the light propagation path and direction change with the random motion of the turbulent medium,which leads to random fluctuation of the focal plane intensity,resulting in image jitter,image blur,image shift,etc.of the observed image phenomenon.Therefore,the research of image restoration algorithm based on deblurring image which effected by atmosphere turbulence is an urgent requirement for high resolution remote sensing image.Based on the summary available atmosphere turbulence degraded image restoration algorithm,according to the application requirements and problems,this paper mainly studies the atmosphere turbulence degraded image restoration and reconstruction,Details are as follows:1.Review and summary on the theory of image restoration.Based on the analysis of remote sensing image degradation model,atmospheric degradation image noise model,PSF(Point Spread Function)prior,a research on the existing classical image restoration algorithm and blind restoration algorithm research findings is investigated,which analysis and comparison on the algorithm principle,characteristics and performance.The image restoration quality evaluation indexes are classified and discussed,and verified against the performance and the advantages and disadvantages of each evaluation index though a practical experimental method.2.An improved blind restoration algorithm based on total variational regularized image is proposed.In this paper,the total variation regularization term is added into the cost function to ensure the quality of the image edge and texture,and the Tikhonov regular constraint term is added to enhance the robustness of the image.In this paper,the Bregman decomposition method is used to decompose the minimization model into several sub-problems,and the spectral decomposition method,the NTRF algorithm and the iterative shrinkage thresholding algorithm are used to solve sub-problem.The experimental results show that the proposed algorithm is superior to the total variational regularization algorithm in image edge and texture preserving ability and image contrast.Objective evaluation indicators show that its clarity index increased by 59%,the average contrast index increased by 24%.3.The MAPEX algorithm of adaptive parameter determination is designed.Based on the prerequisites,algorithm theory and method of solving the APEX algorithm,this paper finds that there are many differences in the OTF(Optical Transfer Function)parameters of the cross-section sampling curve,and the noise interference will cause the OTF parameter estimation deviation Two questions.In order to solve the above problem,the parameter estimation of the single MTF curve is obtained by using the whole MTF(Modulation Transfer Function)surface sampling point fitting parameter estimation.In the case of a given initial value,the empirical value is replaced by the weighted estimate of the support domain sampling point parameter.The experimental results show that the improved MAPEX algorithm has better texture recovery ability and better image recovery than the APEX algorithm.Objective evaluation indicators show that its clarity index increased by 65%,the average contrast index increased by 57%.