摘要
在比较几种椒盐去噪方法的滤波窗口尺寸选择策略的基础上,提出一种基于十字模板的图像椒盐噪声滤除方法.此方法首先将图像中灰度值为0或255的像素点判定为噪声点,然后计算图像中每个噪声点到最近信号点的街区距离,以街区距离为顺序依次对噪声进行均值滤波,滤波运算仅使用以该噪声点为中心的十字模板中的信号点和已处理点.为验证方法有效性,新方法与其他3种椒盐去噪方法相比较.仿真结果表明,新方法与其他方法的最优去噪效果相当,而运算时间明显减少,比获得最优去噪结果的算法快了20~55倍.此外,新方法对现实世界中大量存在的自身包含较多灰度值为0或255像素点的图像,去噪效果优于其他3种方法,而且计算时间较少.
By comparing several policies in selecting the size of filter window for salt and pepper denoising,we propose an approach based on cross template to remove salt and pepper noise in images.The proposed approach firstly considers the pixels of gray level 0 or 255 to be noise pixels,then calculates block distance from each noise pixel to its nearest signal pixel,and sequentially performs mean filtering in order of noise pixel＇s block distance,filtering operation only uses signal pixels and processed pixels in the area of noise pixel centric cross template. Our approach is evaluated by comparing with other three salt and pepper denoising approaches. The simulation results show that the denoising result of the new approach is well-matched with the best result of other three methods,the run time of the new approach is reduced significantly and 20 ~ 55 times faster than the method whose denoising result is optimal. In addition,the denoising result of the new approach is better than other three methods when an image itself contains many extreme pixels of gray level 0 or 255. The new approach can accurately remove the salt and pepper noise,and the run time for the proposed approach is quite short,which makes it possible for the real applications.
出处
《闽江学院学报》
2016年第2期75-81,共7页
Journal of Minjiang University
基金
福建省省属高校专项(JK2014040)
福建省中青年教师教育科研项目(JA15424)
关键词
十字模板
街区距离
椒盐噪声
图像还原
cross template
block distance
salt and pepper noise
image restoration
