基于NLM算法的加权核函数选取研究  

The Selection of Weighted Kernel Function Based on NLM Algorithm

作　　者：徐翠婷 曹剑剑 XU Cui-ting;CAO Jian-jian(College of Computer Science, Jiangxi University of Traditional Chinese Medicine, Nanchang 330004)

机构地区：[1]江西中医药大学计算机学院,南昌330004

出　　处：《现代计算机》2019年第10期68-70,共3页Modern Computer

摘　　要：非局部均值去噪算法利用图像邻域间具有自相似性这一特性,通过加权核函数得到图像中相似图像块的权重,进而达到图像去噪目的。将基于MATLAB平台与传统NLM算法,设计实验令多种加权核函数分别对加高斯白噪声的图像进行去噪处理,再利用图像主观质量评价方法与客观评价方法进行结果分析,一方面可以证明选择合适的加权核函数对于改善去噪效果的必要性,另一方面得到在不同强度高斯白噪声时图像应选择的最佳加权核函数。The non-local mean denoising algorithm utilizes the property of the self-similarity between the image neighborhood, and the weighted kernel function is used to obtain the weight of the similar image blocks in the image, thus achieving the image denoising purpose. In this paper, based on MATLAB platform and the traditional while NLM algorithm, design experiments to make a variety of weighted kernel function to add white gaussian noise, respectively, to deal with the noise of images, reuse subjective image quality assessment method and objective evaluation method to analyze the results, on the one hand can prove that choosing the appropriate weighted kernel function to improve the denoising effect of necessity, on the other hand got in different intensity of Gaussian white noise image should choose the best the weighted kernel function.

关 键 词：图像去噪 加权函数 非局部均值 高斯白噪声 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]