摘要
图像去噪是图像处理领域的一个经典话题,也是一个难点问题。基于邻域小波系数收缩的NeighShrink法比经典的Visu-Shrink法去噪效果要好,但是NeighShrink法在所有的分解层使用次优的通用阈值,致使去噪效果不甚理想。在分析小波系数变化规律的基础上,针对NeighShrink法的不足,提出了一种阈值改进的新方法,该方法具有较好的阈值自适应性。实验表明,该方法正确有效,去噪后的视觉效果得到改善,在客观指标PSNR和MSE上均优于NeighShrink法和经典的VisuShrink法。
Image de-noising is a classical topic and a difficulty in image processing.A method based on the shrinkage of neighboring wavelet coefficients called NeighShrink shows more effective results than VisuShrink,but its disadvantage is to use a suboptimal universal threshold for all sub-bands.In this paper,a new method is presented for threshold selection based on the analysis of variation of wavelet coefficients and NeighShrink.The new threshold is scale adaptive and final experiments prove the availability of the new method,show that it improves the visual effect and outbalanced VisuShrink and NeighShrink on indices PSNR and MSE.
出处
《计算机工程与应用》
CSCD
北大核心
2011年第27期203-205,220,共4页
Computer Engineering and Applications
基金
国家自然科学基金(No.60778051)~~
关键词
图像去噪
小波变换
阈值
邻域小波系数
image de-noising
wavelet transform
threshold
neighboring wavelet coefficients
