基于非局部 CTV-L1模型的大破损彩色纹理图像修复

摘要

In order to solve the inpainting problem for the large broken texture color image,this paper extended TV-L1 model to the non-local CTV-L1 model.Technically,this new model included both non-local operators that was capable of inpainting large broken texture image and CTV regularization term that was able to preserve the edge of color image by taking into consi-deration the coupling of different layers of color images.By introducing auxiliary variables and Lagrange multiplier,it designed the simple augmented Lagrangian algorithm for this model to improve its computational efficiency.Some numerical experiments validate that the proposed model can not only preserve the edge of color image,but also remove anomalies and irregularities from images.The research can be applied to the salt and pepper noise removal of color texture image and color texture image segmentation,etc.%针对大破损彩色纹理图像的修复问题，将 TV-L1模型推广到非局部 CTV-L1模型。该模型不仅包含非局部算子，同时还引入了 CTV（color total variation）规则项，前者可以修复大破损纹理图像，后者充分考虑了彩色图像层与层之间的耦合关系，在处理彩色图像时可以有效地保持边缘。为提高模型的运算效率，通过引入辅助变量和 Lagrange 乘子为其设计了相应的增广 Lagrangian 算法。数值实验结果证实所提出的模型在处理彩色图像时可以有效地保持边缘，同时去除图像中异常的不规则点。该研究可以推广到彩色纹理图像椒盐噪声去除及彩色纹理图像分割中。