一种倒数加权的窗口自适应邻域图像滤波算法

对于叠加了白噪声的图像,提出一种倒数加权的窗口自适应邻域图像滤波算法.算法首先利用自适应邻域统计的概念在以每个滤波点为中心的滤波窗口内,为其建立参与滤波的自适应邻域像素集合,随后对纳入自适应邻域的像素进行倒数加权作为滤波结果.仿真和对比实验的结果显示,提出的算法在性能上超过了传统的中值滤波和窗口自适应邻域滤波算法.它能在有效抑制加性白噪声的同时,完好地保持图像的边界和细节信息,并且滤波后的图像获得了良好的视觉效果.

A Reciprocal Weight Image Filtering Algorithm Using Adaptive-Neighborhood Statistics

An additive noise removing algorithm using reciprocal weight and adaptive neighborhood statistics is proposed. At first, adaptive neighborhood statistics is used to establish adaptive neighborhood of every pixel for filtering. Then reciprocal weight operation is applied to the filtering neighborhood to get the result. Simulation and experiments have been done for images contaminated by various noise level, and the results show that the proposed algorithm outperform the median filter and adaptive-neighborhood filter. It can effectively restrain additive noise without blurring the edge and detail information of images, and also it can provide good visual effect.