改进非局部均值滤波在图像降噪处理中的应用  

Application of Improved Non-local Mean Filtering in Image Denoising

作　　者：戚伟[1] 贾超贤 李洁[3] QI Wei;JIA Chaoxian;LI Jie(School of Intelligent and Electrical Engineering,Huainan Vocational Technical College,Huainan 232001,China;School of Computer Science&Technology,China University of Mining and Technology,Xuzhou 221116,China;School of Computer Science and Engineering,Anhui University of Science&Technology,Huainan 232001,China)

机构地区：[1]淮南职业技术学院智能与电气工程学院,安徽淮南232001 [2]中国矿业大学计算机科学与技术学院,江苏徐州221116 [3]安徽理工大学计算机科学与工程学院,安徽淮南232001

出　　处：《太原学院学报（自然科学版）》2023年第4期59-64,共6页Journal of TaiYuan University:Natural Science Edition

基　　金：安徽省高等学校自然科学研究重点项目(KJ2020A1163)。

摘　　要：针对传统非局部均值滤波应用于图像降噪存在降噪效果欠佳、运行效率低的问题,从两个角度对算法进行改进。采用积分图避免重复性计算,提高算法运行效率。采用相似权重系数融合像素点欧式距离改进相似权重系数与最大特征高斯加权系数,提高算法降噪性能。对比不同算法降噪性能,结果表明改进非局部均值滤波算法不仅对图像噪声平滑彻底,同时还可以有效恢复图像边缘。In response to the problems of poor denoising effect and low operational efficiency in traditional non-local mean filtering applied to image denoising,the algorithm is improved from two perspectives.Integral graphs are used to avoid repetitive calculations and improve algorithmic efficiency.Similar weight coefficients are utilized to fuse pixel Euclidean distance to improve the similarity weight coefficients and maximum feature Gaussian weight coefficients,and improve the algorithmic denoising performance.Comparison results of denoising performance of different algorithms show that the improved non-local mean filtering algorithm not only smoothes image noise thoroughly,but also restores image edges effectively.

关 键 词：图像降噪 非局部均值滤波 积分图 融合相似权重系数 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]