利用图像邻域相关和直方图对椒盐噪声的鲁棒性,提出了一种针对高椒盐噪声图像的软阈值直方图加权滤波算法.对邻域灰度相关进行了量化分析,定义了灰度相关函数作为信号邻域相关性的度量,并将该系数作为直方图加权滤波算法的软阈值,根据像素被判定为噪声或有效信号的概率,自行调整滤波强度,减少图像滤波处理中的细节损失.仿真结果表明,对于高椒盐噪声图像,本算法在椒盐噪声滤除方面有良好的表现.
We proposed a histogram weighted filtering algorithm with soft-threshold for images heavily corrupted by saltpepper impulse noise. It was based on the correlativity principle of image pixel gray. The neighboring correlafivity coefficient was defined to identify noisy pixels and adaptively adjust the filtering intensity. Theoretical and experimental results show that the proposed algorithm has great performances for filtering out impulse noise in extremely corrupted images by high density salt-pepper noise.