一种改进型各向异性扩散滤波器  

NOVEL ANISOTROPIC DIFFUSION FILTER

作　　者：高翔[1] 秦琴[2] 王汝笠[1] 

机构地区：[1]中国科学院上海技术物理研究所,上海20083 [2]上海第二工业大学电子与电气工程学院,上海201209

出　　处：《红外与毫米波学报》2007年第3期237-240,共4页Journal of Infrared and Millimeter Waves

摘　　要：通过分析Perona和Malik(PM)扩散滤波器扩散率函数的统计学意义,提出一种关于扩散率函数的统计学解释模型,即扩散率函数定义了一个以边界估计算子为随机变量的概率密度函数,解决了扩散率函数设计和选择没有统一理论的问题.依据这一解释模型,设计出一种改进型的各向异性扩散滤波器.实验结果显示,改进型扩散滤波器只需要相对于PM方法较少的迭代次数,就能得到预期的图像去噪效果,证明统计学解释模型为各向异性扩散滤波器提供了一种有效的设计方法.The statistic meaning of the diffusivity function of Perona-Malik diffusion filter was analyzed. A statistic interpretation model was proposed that the diffusivity function defines a probability density function （PDF） for the edge estimator as a random variable. This model overcomes the difficulty of no unified theory of designing and selecting the diffusivity functions. A novel anisotropic diffusion filter based on this model was designed. Experiments demonstrate that the novel filter promises the same efficiency in images denoising as PM method, and with less recursion times. So our interpretation model of the diffusivity function provides an effectively method to custom anisotropic diffusion filter for special application.

关 键 词：图像去噪 各向异性扩散 统计学解释模型 

分 类 号：TP991.73[自动化与计算机技术]