基于邻域结构相似性的混合噪音线性滤波算法

A Mixed Noise Linear Filtering Algorithm Based on the Neighborhood Structure Similarity

本文提出一种基于像素邻域结构信息相似性的混合噪音线性滤波算法(GLMF)。该算法是对线性混合滤波器(LMF)的一种改进,它利用图像中存在着大量冗余信息的特性,恢复被混合噪音染污的像素,在判断邻域内像素的相似性时,除考虑像素灰度值的相似性之外,又考虑了像素邻域结构的相似性,用像素灰度值的梯度来表示邻域结构信息。仿真实验证明,用GLMF去噪的视觉效果和峰值信噪比(PSNR)均优于已知的同类滤波器。该算法适用于恢复被高斯噪音和随机脉冲噪音混合污染的数字图像。

：A mixed noise linear filtering algorithm based on the neighborhood structure information＇s similarity （GLMF） is proposed, which is also a good improvement of the linear filter algorithm（LMF）. This algorithm makes full use of the image of redundant information to restore the mixed noise pollution of pixels. In judging the similarity of neighborhood pixels, we consider the pixel values, and consider the similarity of a pixel neighborhood structure, while using the gradient of pixel values to express neighborhood information. Simulation experiments show, comparing with the existing similar filter, GLMF is better on both visual effect and PSNR. This algorithm is applicable to restore the Gaussian noise and random pulse noise mixed pollution digital images.