基于局部统计特性的滤波算法

A New Filter based on Local Statistical Characteristics

提出一种基于局部统计特性的新的滤波算法,可用于去除图像中混合噪音(高斯噪音和一致脉冲噪音),称之为线性局部混合噪音滤波器(简记为LLMF).该滤波器有机地结合了局部滤波算法和线性滤波算法,它采用局部邻域偏差判断像素的相似性,从而提高了对高斯成分的去噪能力.对纯高斯、纯脉冲以及它们的混合噪音都有较好的去噪效果.从实验结果可见,使用LLMF滤波器,图像的视觉效果、峰值信噪比、均方误差均优于同类混合滤波器.

A new filtering algorithm is proposed based on local statistical characteristics in this paper, and it is called Linear Local Mixed Filter （LLMF） which is suitable for digital images corrupted with mixed noises （ impulse noise and Gaussian noise）. The filter combine local filtering algorithm with linear filtering algorithm. It uses local neighborhood deviation to judge similarity of pixel and thus increased the performances of removed Gaussian components. Its denoising results are better to Gaussian pure, pure impulse noise and their mixture. Clearly, It excelled some similar filters in terms of image vision effects, the peak signal to noise ratio, the mean square errors after using LLMF by experiments.