| High-resolution image reconstruction plays an important role in modern life and application,and it applies in important areas such as national defense,medical,security,aviation.For reconstruction of high-resolution image,we should start from the reason of image blurring to solve problems,and if we only consider about hardware requirements to improve the accuracy of instruments,there will be problems of high cost and limited technological level.Therefore,we need to improve effectiveness of algorithms,restore HD images from low resolution images by digital signal processing method.At present,the main idea is to model other reasons that lead to a blurring image the image as a point spread function(PSF)except the noise.Image blurring can be regarded as result of joint function of PSF,HD image convolution and the noise.Achieve the purpose of image reconstruction through accurately estimating PSF,but it's an ill conditioned problem to estimate the PSF.In this paper,estimation of PSF and modeling of algorithm includes the following parts:firstly,we analyze the traditional algorithm of PSF estimation by edge method and find its advantages and disadvantages,then improve it.With the area weighting method,we can effectively estimate the optimal solution of PSF with separation.Find the relationship between blurring image edge and PSF by Radon transform,build the mathematical model and find the optimal solution of PSF with conjugate gradient method.Researches of former scholars are based on the blurring single edge,they don't study the conditions that edge distance is less than PSF diameter.Therefore,we make a deep analysis of this condition in this paper.We build the mathematical model with splicing method for images of which the edge distance belongs to the[r,2r]zones(r is the PSF diameter),for the condition that edge distance is less than r,we make up it with 0.In this paper,we also experimental analyze the influence degree of Radon transformation method by noisy. |
PDF Full Text Request