基于新非线性多重网格法的图像去噪  

Image Denoising Based on New Nonlinear Multigrid Method

作　　者：韩粉 杨奋林[1] HAN Fen;YANG Fenlin(School of Mathematics and Statistics,Jishou University,Jishou 416000,Hunan China)

机构地区：[1]吉首大学数学与统计学院,湖南吉首416000

出　　处：《吉首大学学报（自然科学版）》2022年第1期38-42,共5页Journal of Jishou University(Natural Sciences Edition)

基　　金：吉首大学研究生科研项目(JDY21002)。

摘　　要：通过对差分曲率设置有效的限制算子和插值算子,构造了一种新的非线性多重网格法,并将此方法应用于基于差分曲率的TV^(P)模型.新方法与不动点迭代法的对比实验结果表明,新方法处理的图像峰值信噪比明显高于不动点迭代法,且收敛速度是不动点迭代法的2~3倍.A new nonlinear multigrid(NM)method is constructed by setting effective restriction operator and interpolation operator for differential curvature,and this method is applied to TV^(P) model based on differential curvature.The experimental results show that the peak signal-to-noise ratio(PSNR)of images processed by the new method is significantly higher than that of the fixed point iteration(FP)method,and the convergence speed is 2~3 times that of the fixed point iteration method.

关 键 词：图像去噪 非线性多重网格法 限制算子 插值算子 TV^(P)模型 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]