去除椒盐噪声的非对称有向窗加权均值滤波

Asymmetrical orientation weighted mean filter for salt-and-pepper noise removing

针对传统滤波对称窗口在图像边缘处会引入干扰像素引起图像模糊的问题,提出一种非对称有向窗加权均值滤波算法。首先,基于区域极值进行噪声检测;其次,在对称有向窗的基础上提出非对称有向窗的概念,对于噪声点,通过标准差最小的原则自适应选择非对称有向滤波窗口;然后,在选择的非对称有向滤波窗口内对噪声点进行自适应基于距离倒数的加权均值滤波;最后,根据滤波结果对噪声点进行二次确认,纠正前面噪声检测中误检的噪声点。仿真实验表明,该方法在去噪效果和边缘保持性能及运算速度上都优于基于对称有向窗滤波算法,而且与传统的中值滤波法及其一些改进算法相比,也表现出了更优越的性能。

Because a few interferential pixels will be involved in the filter operation when the traditional filter based on symmetrical window is used,image blurring always occurs in traditional image filtering.This paper presents an asymmetrical orientation weighted mean filter to remove the salt-and-pepper noise in the images.Firstly,the noise pixels are detected based on regional gray extreme value.Secondly,a concept of asymmetrical orientation window is proposed based on symmetrical orientation window,and the actual filter window of noise pixel is selected adaptively according to the principle that the filter window has the minimal standard deviation.Then,a weighted mean filter that the weight value is inverse proportional to the distance to the filter pixel is performed in the selected asymmetrical window.Lastly,a secondary confirmation to the noise pixel is carried out for correcting the wrong-checked noise pixel in forward noise detection.The experimental results show that the proposed algorithm outperforms the algorithms based on symmetrical window in aspects of noise removing,preserving of image edge and processing speed.Comparing with the traditional median filter and some improved algorithms,the algorithm proposed in this paper shows superior capability.