基于改进的字典学习算法的图像去噪方法

提出对基于MOD和K-SVD字典学习算法的图像去噪的两个方面的改进。在字典更新阶段，采用一种新的字典更新方式，在保持支集完备的同时寻找字典和表示法。在稀疏编码阶段，根据前一次追踪过程产生的部分系数进行修正和更新。分别对这两种改进进行了验证，并说明了如何进行更快速的训练以及取得更好的结果，实验结果证实了论文方法的有效性。

Image Denoising Method Based on Modified Dictionary Learning Algorithm

Two improvements of the MOD and K-SVD dictionary learning algorithms are proposed by modifying two main parts of these algorithms ：the dictionary update and the sparse coding stages .A new dictionary-update stage is used to find both the dictionary and the representations while keeping the supports intact .According to the previous sparse-coding , it suggests to update the known representations .These two ideas are tested in practice and show how they lead to faster training and better quality outcome .Experimental results prove the effectiveness of the proposed method .