中文摘要：

针对CT图像的单一色彩性，利用已有的医疗档案图像，提出一种快速非局部均值的图像去噪算法。算法分预处理阶段和实际处理阶段。预处理阶段采用位置敏感性哈希的数据结构建立图像的样本邻域数据库。在实际处理阶段，利用位置敏感性的快速检索结果，采NLM算法进行图像去噪。实验表明，与NLM相比，提出的算法能大大节约计算时间，而且能够有效保存图像边缘和细节。

英文摘要：

A fast non-local means image denoising algorithm is presented based on the single motif of existing computed tomography images in medical archiving systems. The algorithm is carried out in two steps of prepossessing and actual possessing. The sample neighborhood database is created via the data structure of locality sensitive hashing in the prepossessing stage. The CT image noise is removed by non-local means algorithm based on the sample neighborhoods accessed fast by locality sensitive hashing. The experimental results showed that the proposed algorithm could greatly reduce the execution time. as compared to NLM. and effectively preserved the image edges and details.