Research On Image Deconvolution Methods Based On Multichannel Constraints

Image deconvolution has long been a focus of research and timeless topic for scholars with a broader development perspective.Image deconvolution,a degraded image restoration technology,is used to process degraded image to achieve the high-quality image through the corresponding algorithms.However,there are lots of difficulties during the research process because of its own ill-condition and complexity,up to now studies on the image deconvolution have been made a great development,but almost each method has some drawbacks.Therefore,in order to obtain better restoration image,improving the original algorithms and putting forward more effective new methods are especially important.Starting with the image deconvolution,the research situations,significances are introduced and the core methods,key technologies and the problems needing to be solved are mainly analyzed,.Based on the theoretical researches,valuable improvement work are made in the aspects of unknown boundary conditions and the use of the share of information in different channels of a single color image.Several innovative work and research results are concluded as follows:(1)Focusing on the limitations of assumed boundary conditions,a edge inpainting image deconvolution method is proposed through integration of image inpainting and image deconvolution.The method makes deconvolution when the boundary is unknown,which avoids the influence of artificial boundary conditions,and it analyzes the boundary features efficiently during the deblurring.The experimental results show that the proposed method improves the restoration results of boundary,enhances its applicability and proves the excellent speed.(2)Focusing on the lack of information between image channels,a regularization term called Retinex is proposed by making use of the property that the similarity between channels is enhanced after processing by the algorithm Retienx SSR.The regularization term helps different channels constrain each other,by this way,bulrred images can obtain more important prior information with the help of clear image.(3)Focusing on the chromatic aberration,A image deconvolution method based on Retinex prior knowledge is proposed by incorporating(1)and(2)into the deconvolution process,A new image restoration model is bulit,on the one hand,the data fitting term considers the edge inpainting strategy,on the other hand,the regularization term considers the image prior information and a new cross-channel prior(a prior knowledge of Retinex)information,then the extended alternating directions method of multipliers(ADMM)is used to solve the image deconvolution convex optimization problem.In experiments,compared with several state-of-the-art deconvolution algorithms,including BM3 D,HYP,YUV and the third chapter method,the proposed method can obtain a color aberration correction effectively to better cope with degraded images and get high robustness.