计算机与数字工程2017，Vol.45Issue(1)：135-137,155,4.DOI:10.3969/j.issn.1672-9722.2017.01.029
基于Lp范数广义变分正则化图像滤波
Image Restoration Based on Lp Norm Generalized Variational Regularization
摘要
Abstract
The traditional variational PDE methods,such as harmonic model based on L2 norm,total variation(TV) model and the generalized TV model have shortcoming that can not effectively retain edge details during filtering noise.In this paper,a generalized variational filtering method is proposed based on Lp norm model,the differential of curvature is adopted as the adaptive adjustment parameters to realize image anisotropy processing.Simulation results show that the algorithm can effectively detect isolated noise point edge with the flat area,and overcome the staircase effect and preserve the local details.Meanwhile,the proposed method have more obvious advantages both subjective visual effect and objective evaluation than the traditional variational PDE methods.关键词
图像滤波/差曲率/各向异性/Lp范数Key words
image filtering/differential curvature/anisotropy/Lp norm分类
信息技术与安全科学引用本文复制引用
王益艳..基于Lp范数广义变分正则化图像滤波[J].计算机与数字工程,2017,45(1):135-137,155,4.基金项目
四川省教育厅自然科学一般项目(编号:16ZB0355) （编号:16ZB0355）
达州市科技计划应用基础研究项目(编号:KJJ2015001)资助. （编号:KJJ2015001）
