带修正因子的均值和中值滤波算法研究

Research on Methods of Mean and Median Filters with Adjustable Factors

大部分自然图像同时包含高斯噪声和椒盐噪声,简单的运用传统的滤波算法不能得到理想的滤波效果。为了解决混合噪声图像的滤波问题,分别针对以高斯噪声为主的混合噪声图像和以椒盐噪声为主的混合噪声图像,提出了带修正因子的均值滤波算法和带修正因子的中值滤波算法。这两种算法是在修正后的阿尔法均值滤波算法的基础上做了两方面的改进:首先,提出在图像邻域内为不同灰度值的像素点给出归一化的权值,用这些权值和其对应的灰度值共同决定滤波输出。其次,所设计的权值可以用修正因子来进行微调,来获得理想的滤波效果。实验证明,其处理效果优于传统滤波算法和修正后的阿尔法均值滤波算法。

Gauss noise and salt-and-pepper noise exist in many nature images simultaneously.And the effect of traditional mean filter or median filter is not ideal.In allusion to this question,methods of mean filter and median filter with adjustable factors were proposed on the basis of analyzing the character of an amended alpha and mean filter.Both of the two methods introduce two improved aspects：the pixels of different gray values are given normalized weights in the image neighborhood,and the outputs of the filters depend on both of the gray value and the operator weight;The adjustable factor can be used to tune the weight.In real application,the new methods show better performance than traditional filter and amended alpha and mean filter.