Research On The Restoration Algorithm Of Noisy Motion-blurred Images

People get the information,most of which is felt through our eyes in the outside world.And most of information is ultimately retained in the form of images.When we obtain information and store images with different observation equipment and methods,we will be affected by various hardware conditions(such as noise in hardware devices,defocus-blurred,motion-blurred etc.).These factors result in unclear images,even the important information in the image is severely lost,thereby losing its practical value.In order to solve these problems,image restoration algorithms are constantly improved and is widely used in the fields of medicine,aviation,industry and military.But these restoration algorithms have some disadvantages,resulting in poor recovery.So image restoration algorithms need to be continuously studied and improved.In the image restoration algorithm,Because Richardson-Lucy deconvolution algorithm has some advantages:such as it has fast convergence and can better retain image information.It has been used.However,as the number of formula iterations is increased,the algorithm will not only continuously amplify the noise of motion-blurred with noise image,but It can even cause serious distortion and distortion of the image.In order to solve the shortcomings of the traditional image restoration algorithm,and preserve the image high frequency information better.Two new image restoration algorithms are proposed.Which is based on empirical mode decomposition combined with R-L algorithm;that is based on adaptive median filtering combined with R-L algorithm.Both algorithms improve image clarity and image quality.The main work of the dissertation is as follows:1.Firstly,the formation and analysis methods of motion-blurred with images are introduced from five aspects:degenerate model of image,types of noise,image denoising method,image restoration algorithm,image quality evaluation and so on.At the same time,the advantages and disadvantages of several image denoising methods and image restoration algorithms are analyzed.2.An image restoration algorithm based on a combination of Richardson-Lucy and adaptive median filtering is proposed.This algorithm is compared with R-L algorithm and image restoration algorithm based on R-L and median filtering.And the principle and calculation process of the algorithm are given.The algorithm was simulated in terms of subjective evaluation of the image,peak signal-to-noise ratio,and time-consuming.By comparing three algorithms,it is proved that this method can effectively suppress the noise amplification problem in the image,and it can enrich better the edge information of the image,and obtain a better image restoration effect.3.When image algorithms are processed,Because the high-frequency information and noise information of the image are close to each other,the high-frequency information is filtered out as noise in the process of noise filtering,thus losing some of the information of the image.The resulting image features are not obvious.Therefore,in order to overcome this defect,two-dimensional empirical mode decomposition and principal component analysis are introduced into the algorithm.Firstly motion-blurred image with noise is decomposed into high-frequency layer and low-frequency information layer using two-dimensional empirical mode.Then the high-frequency layer containing the noise information is determined by principal component analysis method to analyze the number of filter layers.When noise is filtered out,high frequency information is retained better by the algorithm.After comparing with three algorithms,it is proved that this method preserves the image features well,can enrich the details of the images better,and recovery quality of the image has been obviously improved.This article mainly studies image restoration of motion-blurred image with noise,two kinds of image restoration algorithms are proposed.The first algorithm solves the Richardson-Lucy algorithm disadvantages of noise amplification and results instability.But,there is improved in retaining some information of the image.Therefore,in order to suppress this defect,a new algorithm is proposed.The algorithm preserves the high-frequency information of the image better,can further enriches the details of the image,and has a certain significance for image restoration.