Analysis Of Degraded Image Restoration Results And Research On Evaluation Algorithms

In the modern society,with the rapid development of information,a large part of the information is obtained from the image.How to obtain and get the high-quality image becomes more and more important.However,in the process of optical imaging,image degradation is caused by the defects of optical system,atmospheric turbulence and the rapid motion of the target object,so the image restoration algorithm emerges as the times require.Image restoration is an image processing technique which uses the prior information of image degradation to establish the image degradation model and to restore and reconstruct the original image accurately.In the research of this paper,we mainly consider image denoising and image de-blurring two kinds of image restoration problems,and the ring effect caused by the restoration is evaluated and analyzed.This paper first introduces the factors affecting image quality.(Blur,noise)then the defect of the existing universal image restoration algorithm is analyzed by combining the theory and experiment.As the ringing effect will inevitably occur in the process of image restoration,a method for evaluating the ringing effect of the restoration image is proposed based on the physical mechanism of the ringing.Methods the region of overshoot and ripple of ringing image was extracted,and the gradient of the region was weighted.Through many experiments,the weight value is set to represent the weight of detail information on the edge,and the severity of ringing effect is quantified.This method can effectively evaluate the ringing effect in restoration images with different restoration algorithms and different restoration parameters.The restoration effect of different scenes and degraded degree is studied by simulation experiment.Selecting images of different scenarios from TID2008 and LIVE databases(Characters,scenery,fingerprints,remote sensing,and so on),Adding different degrees of blur and noise to each scene to get the degraded image.The restoration algorithm is used to process the degraded image and the quality evaluation algorithm is used to obtain the noise,blur and ring parameters.Finally,according to the evaluation results,the effect of various restoration algorithms on fuzzy and noise processing and the severity of ring effect caused by restoration are analyzed,and the applicable scenarios of different algorithms are analyzed.