Research On Image Inpainting Based On The Total Variation Algorithm

Data image inpainting is to deal with image, photo or film, and let the injured pattern was renewed and removed or replaced the objects chosen in the picture, which is an important branch of data image restoring technology. The work principle of it is to use the former and last information of the field or a series of injuring image, which is in the field of awaiting data image to fill image' s injured field or awaiting mending field. The main purpose of data image inpainting is let the observers could not observe that images have been corrected and get a better sense of sight. It mainly been used in the coding, transferring, compressing, or inpainting injured images, hiding aims and so on. With the development of computer, how to finish image inpainting automate or half-automate on the base of data image, becomes an important task in the field of image inpainting or research.The apply in image inpainting of the total variation draws our great interest after American Rudin used it in the research of image restoration, but about its research we only focus on the removing noise , involve in image inpainting only a little. How to use its characteristics and let it serve for image inpainting, is the main content of this paper.I firstly discuss on relevant theories of total variation, let it in the use of removing noise and keeping image part edge, at the same time with the characteristics of empty awaiting inpainting field of narrow strip, to improve it and to mend the awaiting field with the two sides information of awaiting inpainting. It' s proved by many experiments that improved total variation is fit for inpainting awaiting image and inpainting of the narrow and empty field in the awaiting inpainting , we are very satisfied with its effects.Except better effect is proved, I do a deeper research on its apply field at the last chapter. It is be proved that it is difficult to inpainting image with it when more awaiting inpainting field edge or larger of it appears, so the use of it in both two kinds of image on the base of relevant board is studied, and it' s proved that it' s efficacious, by comparing total variation with board matching, we know the differencebetween awaiting inpainting and awaiting inpainting image field, and there is a great difference of apply field of various methods. How to use the characteristic of them better, and let them serve for image inpainting, is a worthy research.