针对非局部均值中相似度的衡量问题，提出了一种基于字典学习的度量算法。-For non-local means to measure the similarity of the problem, and proposes a measure of dictionary-based learning algorithms.