基于模糊神经元网络的改进型中值滤波器

Improved Median Filter Based on Fuzzy Neuron Network

本文提出了一种新颖的基于模糊神经元网络的改进型中值滤波器。该滤波器由两部分组成 :第一部分是噪声检测器 ,将待处理像素与其八邻点的灰度差值送入训练好的模糊神经元网络 ,检测出噪声像素 ;第二部分是中值滤波器 ,对于检测出的噪声像素 ,用其相应窗口内的非噪声像素的中值进行替代。实验结果表明 ,本文提出的滤波方法既能有效地滤除噪声 ,又可以很好地保护图像的细节。

In this paper, a novel improved median filter based on fuzzy neuron network is proposed. The proposed filter is composed of two subunits. The first subunit is noise detector, which aims at detecting noise pulses based on trained fuzzy neuron network by considering luminance differences among neighboring pixels. The second subunit is median filter, which use the median of non-noise pixel in the window to replace the detected noise pixels. Experimental results show that the proposed method can efficiently remove noise while preserve image details very well.