基于曲波域Mapshrink算法的双阈值去噪  

DOUBLE-THRESHOLD DENOISING BY APPLYING MAPSHRINK TO CURVELET DOMAIN

作　　者：杨永锋[1] 李红松[1] 周浩[1] 王玲[1] 朱丽青[1] 

机构地区：[1]云南大学信息科学与工程学院

出　　处：《计算机应用与软件》2011年第3期121-124,共4页Computer Applications and Software

基　　金：云南省教育厅基金项目(08Y0044);云南大学校级基金项目(2005Q015C)

摘　　要：Mapshrink算法广泛应用于小波域的阈值估计,而曲波变换是一种新的多方向性的多尺度变换理论。首先将Mapshrink算法引入曲波域,并分析单一阈值去噪存在的问题,提出一种联合全局阈值和局部自适应阈值的双阈值去噪方法。方法既继承了全局阈值保持图像总体灰度的特性,又保留了局部自适应方法保持边缘细节的优点。实验证明,该方法比单纯在曲波域中使用全局阈值或局部自适应阈值有更加显著的效果。Mapshrink algorithm is widely used for threshold estimation in wavelet domain,and Curvelet transform is a new multi-scale transform theory with multi-direction.In this paper,we first apply Mapshrink to Curvelet domain,and analyse the existing problem in single threshold de-noising method,then propose a de-noising method with dual threshold,i.e.combining the global threshold with local adaptive threshold.This method succeeds the characteristic of global threshold in keeping overall gray-scale of the image and also preserves the advantage of local self-adaptive method in maintaining edge details.Based on the experimental results,this method has more remarkable effect than the de-noising methods of global threshold or local adaptive threshold in Curvelet domain.

关 键 词：曲波变换 全局阈值 局部自适应阈值 最大后验估计 双阈值 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]