基于贝叶斯双变量模型和Contourlet变换相结合的红外图像去噪  

A Denoising Algorithm Based on the Combination of Bayesian Bivariate Model and Contourlet Transform

作　　者：杭丹萍[1] 梁栋[1] 马雪亮[1] 韦卫东[2] 唐王琴[1] 徐慧[1] 

机构地区：[1]安徽大学计算智能与信号处理教育部重点实验室,安徽合肥230039 [2]解放军电子工程学院,安徽合肥230037

出　　处：《红外技术》2010年第10期591-594,600,共5页Infrared Technology

基　　金：国家自然科学基金项目;编号:60772121;安徽大学211工程学术创新团队基金项目

摘　　要：提出了一种基于贝叶斯双变量模型(Bayesian Bivariate Model)和Contourlet变换相结合的红外图像去噪算法。首先对含有加性高斯白噪声污染的红外图像进行Contourlet变换,得到各尺度各方向上的Contourlet系数;然后用贝叶斯双变量模型去挖掘图像Contourlet系数的尺度间相关性;最后对处理后的系数进行Contourlet反变换重构,得到去噪后的图像。实验结果表明,该方法有效地捕获了红外图像的轮廓信息,提高了图像的峰值信噪比,改善了图像的视觉效果。Based on the combination of Bayesian Bivariate Model and Contourlet Transform,an algorithm for infrared image denoising is proposed.Firstly,in order to get Contourlet coefficients in all scales and directions,infrared image with additive white Gaussian noise is processed by Contourlet Transform.Then,Bayesian Bivariate Model is used to exploit the dependencies of the Contourlet coefficients across the scales.Finally,we perform inverse Contourlet Transform to the processed coefficients and get the denoised image.The experimental results demonstrate that the proposed method not only captures the contour information of infrared images more effectively,but also improves Peak Ratio of Signal to Noise and visual effects of the image remarkably.

关 键 词：贝叶斯双变量模型 CONTOURLET变换 红外图像 去噪 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]