摘要
针对非局部均值算法对边缘去噪能力较差的缺陷,提出了一种新的自适应非局部均值去噪算法。一方面利用基于均方误差最小化准则的主动匹配,以确定两图像块之间的最佳匹配形状与尺寸,进而得到较为鲁棒的相似度估计。另一方面,利用局部Hessian矩阵特征值判断图像块类型,并据此进行滤波窗口尺寸的自适应调整。细致分析与仿真结果表明,新算法有效克服了原始非局部均值算法存在的边缘去噪能力较差的问题,综合去噪性能达到甚至超过了最新的自适应非局部均值算法。
In order to improve the denoising performance in edge area,a novel adaptive nonlocal means image denoising algorithm is proposed. First,by active matching based on minimum mean squared error criterion,the optimum matching shape and size between two image patch are obtained,leading to more robust similarity estimate. Second,based on the patch type,which is determined through eigenvalues of local Hessian matrix,the size of filtering window is automatically adjusted to adapt to local image characteristics. Systematic analysis and simulation results show that the proposed method has overcome some major drawbacks of original nonlocal means algorithm,and arrived at comparable or superior performance with respect to the latest state of the art modified adaptive nonlocal means denoising methods.
出处
《南阳理工学院学报》
2015年第2期58-63 68,68,共7页
Journal of Nanyang Institute of Technology
基金
国家自然科学青年基金项目(60802047)
