基于置信区间的自适应加权均值滤波算法  

Adaptive weighted mean filtering algorithm based on confidence interval

作　　者：陈家益 黄楠[2] 熊刚强 曹会英 徐秋燕[3] 

机构地区：[1]广东医科大学信息工程学院,广东湛江524023 [2]南京理工大学理学院,江苏南京210094 [3]湛江中心人民医院外科重症加强护理病房,广东湛江524037

出　　处：《南京理工大学学报》2017年第3期307-312,共6页Journal of Nanjing University of Science and Technology

基　　金：国家自然科学基金(61170320);广东省自然科学基金(2015A030310178)

摘　　要：为了改善图像滤波的效果,提出1种基于置信区间的自适应加权均值滤波算法。根据高斯噪声的特点以及其对原图像的影响,仅对滤波窗口中处于置信区间的像素求加权均值。同时考虑了灰度相关性与距离相关性,将灰度测度因子和距离测度因子进行线性加权求和,得出加权系数。最后对加权均值滤波后的图像进行折中的灰度均衡化。实验结果证明,相对于标准均值滤波(SMF)算法和自适应均值滤波(AMF)算法,该文算法的滤波图像更加清晰,很好地恢复了原图像,同时保留了图像的边缘和细节;该文算法对应的归一化均方误差明显低于SMF算法和AMF算法。An adaptive weighted mean filtering algorithm based on a confidence interval is proposed to improve the results of filtered images. The weighted means of the pixels in a filtering window and within the confidence interval are calculated according to the characteristics of Gaussian noise and its effect on an original image. A weighted coefficient is obtained by the linear weighted sum of the gray measure factor and distance measure factor, and the gray correlation and distance correlation are taken into consideration. Finally,the gray of the weighted mean filtered image is equalized. The ex-perimental results show that this algorithm is better than the standard mean filtering（ SMF） algorithm and adaptive mean filtering （ AMF） algorithm, the filtered image is clearer, the original image is recovered well,and the edges and details are kept;the normalized mean square error（NMSE） of this algorithm is lower than that of the SMF and AMF.

关 键 词：置信区间 均值滤波算法 自适应滤波算法 高斯噪声 灰度相关性 距离相关性 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]