Research On Image Restoration Based On Hierarchy Priori Learning

With the entry of society into the era of full information,image restoration technology has been widely concerned,and the method based on transcendental information learning plays an increasingly important role.Compared with traditional restoration methods,this kind of method involves wider range of information and better applicability,and often achieves relatively good restoration results.The improvement of this kind of method often contains two key problems.The first is how to get a more comprehensive and sufficient prior knowledge from the image,and the second is how to use the optimized algorithm to get better image restoration effect.This article mainly carries out the research from the two problems.Image not only contains the statistical properties,and also contains many geometric structure information.It is not enough to describe the attributes of the image accurately and completely by using a single attribute prior which will lose a lot of important image properties.For this reason,we first propose a method of image restoration based on multi-layer prior learning to remove the noise and retain more details,such as edges and textures.We consider that the guided image filtering has better noise suppression effect in homogeneous region,whereas its edge-preserving property is poor.The key to guided filtering is the selection of the guide image.The result of EPLL denoising helps achieve a good structure,but it is easy to cause the ladder effect in homogeneous areas.For this reason,we propose a method of coupling EPLL and guided filtering for image denoising that incorporates these two methods,which complement and promote each other,thus enhancing the effect of image denoising.The research work carried out in this paper mainly includes the following parts:(1)We propose a multi-layer prior information learning method,which combines the statistical and geometric features of the image.By utilizing different image attributes of different layers,and modifying the prior learning classification process under the guidance of this property,this method obtains a more comprehensive and accurate prior dictionary for subsequent image restoration process.(2)Based on the complementarity of EPLL with guided filtering,we propose a method of coupling the EPLL and guided filtering for image denoising.It uses EPLL model to construct the guided image for the guided filtering,which can provide better structural information for the guided filtering.Meanwhile,by the secondary smoothing of the guided image filtering in the image homogenization areas,we can improve the noise suppression effect in those areas,and reduce the ladder effect brought by EPLL.