改进的脊波变换图像半软阈值降噪方法  

Image Semi-soft Threshold Denoising Method Based on Improved Ridgelet Transform

作　　者：罗忠亮[1,2] 林土胜[2] 

机构地区：[1]韶关学院电子与通信工程系,韶关512005 [2]华南理工大学电子与信息学院,广州510641

出　　处：《计算机科学》2009年第3期241-243,共3页Computer Science

基　　金：国家自然科学基金项目(60472006);广东省自然科学基金团队研究项目(04205783)资助

摘　　要：脊波变换是一种源于小波又高于小波的多尺度几何分析方法,应用于图像中。借鉴小波去噪的思想提出一种新的图像去噪方法,采用基于Bayesian估计的自适应阈值和半软阈值技术进行去噪,针对脊波变换所产生的轻微的"划痕",引入平移不变的方法消除这种条纹干扰。实验结果表明,该方法较好地处理了图像细节和边沿保留与噪声抑制的矛盾,是一种有效的去噪方法。Ridgelet is a multi-scale geometric analysis method that comes from wavelet and is better than it,which is applied in the image processing. This paper presented a new denoising way using the essence of wavelet image denoising, the new one is based on adaptive threshold of Bayesian estimation and semi-soft threshold techniques. In view of little scratch caused by ridgelet transform, the interference was eliminated with the method of shift invariant. The experimental results show that this method can better deal with contradictions between image details and edge reservations and noise suPPression.it is an effective denoising method.

关 键 词：半软阈值去噪 脊波变换 平移不变 贝叶斯估计 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP303[自动化与计算机技术—计算机科学与技术]