一种增强结构信息的图像平滑算法  

An Image Smoothing Method with Structure Information Enhancing

作　　者：张道兵[1] 张慧[2] 孙显[2] 刘波[1] 王宏琦[2] 

机构地区：[1]中国科学院西安光学精密机械研究所,西安710119 [2]中国科学院电子学研究所

出　　处：《光子学报》2007年第B06期303-308,共6页Acta Photonica Sinica

摘　　要：基于非线性扩散滤波提出了可以同时对同质区域进行平滑和对结构信息进行增强的算法.该算法在同质区域采用各项同性滤波而在边缘附近采用各项异性扩散滤波,避免了图像平滑和增强结构信息之间的矛盾.在实现时,根据一致性程度选择扩散矩阵的扩散系数,可确保偏微分方程在离散化时的简洁性和一致性.数字实验结果表明,可以有效的平滑一致性低的区域并增强一致性高的结构.It presents a numerical scheme for smoothing homogeneity with coherence enhanced based on nonlinear diffusion filter. Isotropic diffusion in homogenous areas and anisotropic diffusion in edge neighbors are uniformed into the scheme, which could avoid the paradox of smoothing homogenous areas and enhancing structure information. During implementing, switching diffusion coefficients in diffusion matrix according to coherence ensures the discretization of partial differential equation to be concision and consistency. Numerical examples show that it is efficient in smoothing low coherence areas and enhancing coherence structures.

关 键 词：图像处理 图像平滑 非线性扩散 一致性增强 结构张量 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]