自适应正则化项去除乘性噪声  

Removal of Multiplicative Noise by the Adaptive Fidelity Term

作　　者：赵志龙[1] 刘卉[2] 张欢[1] 

机构地区：[1]西安电子科技大学理学院,陕西西安710071 [2]东北大学理学院,辽宁沈阳110004

出　　处：《电子科技》2011年第1期1-3,8,共4页Electronic Science and Technology

基　　金：国家自然科学基金资助项目(60872138)

摘　　要：针对乘性噪声去噪算法研究,乘性噪声多出现在合成孔径雷达、超声波和激光等相干图像系统中,与标准高斯加性噪声不同,乘性噪声符合瑞利和伽马分布函数。文中通过:(1)取对数把乘性噪声模型转变成相加形模型;(2)改进正则化项成为自适应扩散模型;(3)将数学模型应用于图像处理的实践当中。在解决了阶梯效应的同时保持了图像的边缘。Multiplicative noise, unlike the standard Gaussian additive noise, often appears in synthetic aper- ture radar and sonar and laser imaging. The muhiplicative noise agrees with Rayleigh and Gamma functions. This paper converts the multiplicative model into an additive one by taking logarithms, changes the standard TV regularizer into an adaptive fidelity term diffusion model and applies mathematical models into multiplicative imaging denoising. The model can not only remove noise but also enhance edges.

关 键 词：乘性噪声 图像恢复 总变分 自适应正则化项 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]