基于阿尔法均值滤波和信息散度的图像滤波研究  

Filter of the Image Based on Alpha-Trimmed Mean Filter and Information Divergence

作　　者：常丽萍[1] 李伽[1] 施朝霞[1] 郭淑琴[1] 

机构地区：[1]浙江工业大学信息工程学院,浙江杭州310014

出　　处：《光学学报》2014年第B12期190-195,共6页Acta Optica Sinica

基　　金：国家自然科学基金（61304124,61205121,61306090）、浙江省自然科学基金（LYl3F010009）、国家博士后基金（2012M311386）

摘　　要：光学图像感染噪声是一种非常常见的现象，为了克服传统图像滤波方法不可避免带来的图像边缘模糊和奇异值影响，提出了一种基于阿尔法均值滤波算法和信息散度的图像自适应滤波算法。该算法充分利用了信息散度的特性和阿尔法均值滤波算法的优点，采用信息散度自适应的确定滤波窗口的加权系数值。实验结果表明，该算法可动态的确定滤波窗口内像素点的取值，具有较好的自适应性，同时与现有的中值滤波和均值滤波算法相比，对于受到高斯噪声、椒盐噪声以及混合噪声感染的图像，具有比较好的滤波效果，并可以很好地保护图像细节信息，具有提高图像清晰度的功能。The optical image disturbed by noise is a very common phenomenon. In order to overcome the fuzzy effects among the imaging edges and those of singular values induced by the traditional image filtering methods, an adaptive filtering algorithm is proposed on the basis of Alpha-trimmed averaging filter and information divergence theorem. The algorithm makes full use of the advantages of the Alpha-trimmed averaging filter and information divergence. It can adaptively adjust the filter window′s coefficients according to the information divergence. The experimental results show that the high adaptive characteristic due to the dynamically determination for the pixel value in the filtering window. Meanwhile, for the image corrupted by Gaussian noise, salt and pepper noise or mixed noise, the proposed algorithm not only accounts for much more advantages in filtering capacities, but also enhances the detailed distinguishing features of images and the corresponding resolutions while comparing with the present middle and average ranged filtering algorithms, respectively.

关 键 词：图像处理 阿尔法均值滤波 信息散度 

分 类 号：TN911.73[电子电信—通信与信息系统]