一种基于PCNN的有效去除高斯噪声的方法  

An effective method of image Gauss noise filtering based on PCNN

作　　者：刘莉[1] 谈文蓉[1] 

机构地区：[1]西南民族大学计算机科学与技术学院,成都610041

出　　处：《西南民族大学学报（自然科学版）》2012年第4期642-647,共6页Journal of Southwest Minzu University(Natural Science Edition)

摘　　要：对传统的PCNN模型进行了简化,对高斯噪声的特性进行了分析,提出了基于PCNN模型、维纳滤波、均值滤波组合的方法用于去除高斯噪声,该方法不仅能有效地去除噪声,同时能更好地保持了图像细节和边缘.计算机仿真实验结果表明,使用该方法可有效地恢复被噪声污染的图像.且恢复图像的信噪比增量高于用另两种常用的图像恢复方法(中值滤波与均值滤波)得到的结果.The principle of pulse coupled neural network（PCNN） is disussed, and the PCNN model is properly simplified. A new mothod that combines the PCNN model with the Weiner filter and the mean filter is proposed for the removal of gauss noise in digital images. Compared with other traditional filtering methods, this method works well, and has strong adaptability as well as protecting image edge and image detail while noises are decreased. The experiment results and comparisons show that PSNRs of restored images by using PCNN are higher than those by using one of two usual image restoration methods, the median filter and the mean filter.

关 键 词：脉冲耦合神经网络 维纳滤波 均值滤波 高斯噪声滤波 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]