基于直方图的自适应图像去噪滤波器

Histogram-Based Adaptive Filter for Removing Image Noise

对于那些明显偏离高斯型白噪声的加性噪声 ,如拖尾脉冲噪声 ,高斯脉冲噪声等 ,已有方法的滤噪性能会严重退化 .为此 ,该文提出了一种去除脉冲噪声的新方法 .该方法首先由被污染图像估计出原图像的直方图 .然后应用模糊集理论 ,利用加权策略得到了一个符合图像灰度分布统计规律的模糊隶属度函数 ,以此隶属度函数构建一个加权平均滤波器 .新方法有效地利用了原图像的先验知识 ,能够根据图像区域特性差异及脉冲噪声强弱自适应地采用不同的滤波尺度 .文章比较了传统滤波器、已有的模糊滤波器和本文方法的结果 .

The performance of the existing filters will badly deteriorate in removing additive impulse noise, such as Gauss impulse noise and long-tailed impulse noise. Therefore, a new method for image restoration is presented. The new method can be used to estimate histogram of original image through input image, and gets a membership through this histogram by using fuzzy set theory, then establishes a weighted fuzzy mean filter through this membership. The new method effectually utilizes the prior knowledge of the original image, and can adaptively adopt different filter scale in the light of character divergence of image region and intensity of impulsive noise. The performance of the new method is compared with conventional filters and available fuzzy filters. Experimental result shows that the performance of the new method is better.