期刊名称：International Journal of Innovative Research in Computer and Communication Engineering
印刷版ISSN：2320-9798
电子版ISSN：2320-9801
出版年度：2017
卷号：5
期号：5
页码：9660
DOI：10.15680/IJIRCCE.2017.0505069
出版社：S&S Publications
摘要：An increasingly popular area of research in the field of digital image processing is image inpaintingwhere the removal of undesired objects and reconstruction of the missing or damaged parts of the digital images hasbecome the fundamental problem. Image inpainting is the practice of solving these problems where the damaged ormissing portions of the images are reconstructed in a visually plausible manner. In this paper, an effectual inpaintingtechnique is presented for the removal of the unsought objects and reconstruction of the digital images. The basic ideabehind this technique is to convert the multichannel image into multiple-bands (RGB) and then employ BOS andNLTV model for the optimization of proposed multiband-MNLTV model. The proposed model is tested over a numberof digital images. The experimental results prove the multiband-MNLTV model a novel approach, in both its accuracyand contrast invariance.
关键词：Inpainting; multiband; multichannel nonlocal total variation (MNLTV); image restoration
