基于UDCT的改进双变量模型图像去噪  

Image denoising with improved bivariate model based on uniform discrete curvelet transform

作　　者：杨兴明[1] 牛坡礼 

机构地区：[1]合肥工业大学计算机与信息学院,安徽合肥230009

出　　处：《合肥工业大学学报（自然科学版）》2015年第3期345-350,共6页Journal of Hefei University of Technology：Natural Science

基　　金：安徽省自然科学基金资助项目(090412041)

摘　　要：文章通过对均匀离散曲波变换(UDCT)域中小波系数统计特性的研究,针对传统双变量模型未考虑空间聚集性的不足,提出了一种新的双变量模型去噪算法。首先在双变量模型的基础上采用了蒙特卡洛方法估计各子带的噪声方差;然后引入邻域模型,通过调整邻域窗的大小估计相应窗口内小波系数的度量方差,得到初始化图像;最后以初始化图像和原噪声图像为先验信息,推导出改进的双变量模型来处理原噪声图像,且以对称K-L散度和最大迭代次数为收敛条件,得到最终去噪图像。实验结果证明了该算法的有效性。By researching on the statistical properties of wavelet coefficients of the uniform discrete curvelet transform（UDCT）domain,a new denoising algorithm based on the bivariate model is proposed to make up for the unconsideration of spatial clustering in traditional bivariate model.Firstly,the Monte Carlo method is used to estimate the noise variance of each subband on the basis of the bivariate model.Secondly,in order to obtain the initial image,the neighborhood model is introduced to estimate the measurement variance of the wavelet coefficients of corresponding window by adjusting the size of neighborhood window.Finally,the initial image and the original noising image are used as prior information to deduce the improved bivariate model for dealing with the original noising image.And the final denoising image is acquired under the convergence condition of symmetric Kullback-Leibler divergence and the maximum number of iterations.The experimental results prove the effectiveness of this method.

关 键 词：均匀离散曲波变换 蒙特卡洛方法 邻域模型 双变量模型 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]