Research On The Algorithms Of Digital Image Inpainting

Image inpainting is an aneient art.In renaissance of Europe,People have repaired the cracks of the work of fine art.They can resume the damaged portions of a work of fine art and keep it whole by this way. Its objective is to restore the missing or damaged portions of the image inorder to make it more legible and to restore its unity in away that is non-detectable for an observer who does not know the original image. This technique is used widely in many fields such as repairing damaged medical image or ancient artifact,inpainting injured photo,removing extra object in image,restoring scene of the crime and so on.Currently the main mehtods for image inpainting include texutre synthesis algorithms and the algorithms based on partial dieffreniial equations.Researching status of image inpainting is summarized and some representative inpainting algorithms and their advantages and disadvantages are analyzed in this paper.On this basis, three inpainting algorithms are proposed.Analysis the image inpainting algorithm based on total variation (TV) model. The inpainting algorithm of TV model using the anisotropic diffusion only, anisotropic diffusion of TV model just diffuses along the image edge and introduces step effect easy in smooth regional. Now an improved image inpainting algorithm is proposed. Improved algorithm combines isotropic and anisotropic diffusion and uses the regional frequency difference to realize different iterative equation on the different regional. The step effect in original algorithm can be effective avoided and raise iterative efficiency on smooth regional.Propose a new algorithm based on texture synthesis that uses a new priority function to improve some deficiency in calculating the patch priority in Criminisi algorithm. Take variance to represent frequency for template selection to inhibit the noise's effect for template selection and reflects the local difference more better to search the optimum matching block more accurately. The algorithm has been verified its accuracy during the experiments.Considering the limitation of Total Variation model for image impainting. An improved inpainting model based on image decomposition is proposed. Using image decomposition technology, extract the structure information and texture information from the image. The improved TV model are applied to structure part of image to effectively avoid step effect in TV model on smooth regional. On the process of iterative, respectively analysis the feature points and non-feature points to avoid the fuzziness of feature points during the inpainting. Take improved exemplar-based inpainting technology for the image texture information.