一种基于相似度与改进高斯函数的滤波新方法

A new filtering algorithm based on similarity and Gaussian function

为了解决低密度噪声对图像的影响问题,提出了一种基于相似度与高斯函数的新滤波算法.该算法根据相似度对噪声点进行检测,通过引入正确率和召回率两个指标动态地调整相似度函数的参数和噪声检测时的阈值.针对传统噪声滤波器存在的问题,通过调整传统高斯滤波器的中心权重和方向性,使得滤波器能够根据图像局部区域的相似度特征,生成自适应的各向异性高斯滤波模板,对检测到的噪声点进行滤除.仿真实验表明,在低密度噪声污染下,该算法在有效抑制噪声的同时能够更好地保留边缘细节信息.

In order to solve the influence of low-density noise on images,a new filtering algorithm based on similarity and Gaussian function was proposed.The algorithm detected noise points according to similarity,and adjusted the parameters of similarity function and threshold value of noise detection dynamically by introducing accuracy and recall.In view of the problems existing in traditional noise filter,the new adaptive anisotropic Gaussian filter can adjust the center weight and direction according to the similarity characteristics of the local area of the image.Experimental results showed that the algorithm can effectively suppress noise and retain edge details better than other algorithms under low-density noise pollution.