中文说明：信号的稀疏表示受到相当大的近几年的利息。假设那自然信号，如图像、 承认稀疏分解在冗余字典会导致处理这类消息来源的有效算法数据。尤其是，设计的很好适应词典图像一直是一项重大挑战。最近有 K 和奇异值分解提出了这项任务并显示要很好地执行各种灰度图像处理的任务。在本文中，我们处理图像和扩展学习词典颜色问题基于 K 奇异值分解的灰度图像出现的去噪算法在中。这工作提出处理非齐次噪声和缺少的信息，为国家艺术铺平道路在应用程序 （如彩色图像去噪，马赛克，结果和修复，如本文所示。
English Description：
Sparse representations of signals have drawn considerableinterest in recent years. The assumption that natural signals,such as images, admit a sparse decomposition over a redundantdictionary leads to efficient algorithms for handling such sourcesof data. In particular, the design of well adapted dictionaries forimages has been a major challenge. The K-SVD has been recentlyproposed for this task and shown to perform very well for variousgrayscale image processing tasks. In this paper, we address theproblem of learning dictionaries for color images and extend theK-SVD-based grayscale image denoising algorithm that appearsin. This work puts forward ways for handling nonhomogeneousnoise and missing information, paving the way to state-of-the-artresults in applications such as color image denoising, demosaicing,and inpainting, as demonstrated in this paper.