梯度相似性的椒盐图像加权中值滤波算法  

Weighted Median Filtering of Im Based on Grads Similarity

作　　者：李迅波[1] 蒋东升[1] 王振林[1] 

机构地区：[1]电子科技大学机械电子工程学院,成都611731

出　　处：《电子科技大学学报》2012年第1期114-119,共6页Journal of University of Electronic Science and Technology of China

基　　金：四川省科技厅基金(2010JY0053)

摘　　要：在对现有椒盐噪声中值滤波算法分析的基础上,提出了基于梯度相似性的椒盐噪声图像加权中值滤波算法。利用灰度图像窗口内各个像素点灰度值的差异,将含有椒盐噪声的图像分为疑似噪声点和信号点,然后利用窗口像素点的梯度相似性对疑似噪声点进行分析,并运用图像极值剪切技术去除噪声点像素的干扰。采用舍弃方差极大值的改进加权中值滤波算法给噪声点赋值。实验表明,该算法对图像的细节保留能力和滤波能力有较大的提高,能取得较好的峰值信噪比。Through analyzing existing methods of median filtering on salt pepper noise,a new improved weighted median filtering based on grads similarity is proposed.The proposed algorithm divides the image with salt pepper noise into uncertain noise pixel and signal pixel from the difference of pixel grey value in filtering window.The uncertain noise pixels are further analyzed by grads similarity in filtering window.The disturbance of noise pixel is removed by applying extremum trimming method.At last,a improved weighted median filtering algorithm with abandon points of max square deviation is used to compute the grey value of noisy point.Simulation results show that this proposed method has a great improvement in removing salt pepper noise and preserving the details of image.The restored image can also obtain better peak signal to noise ratio（PSNR）.

关 键 词：极值剪切 梯度相似性 峰值信噪比 椒盐噪声 加权中值滤波 

分 类 号：TP317[自动化与计算机技术—计算机软件与理论]