基于改进PCNN决策的非对称裁剪中值去噪方法  

Unsymmetric Trimmed Median Filter Method Based on Improved PCNN Decision

作　　者：朱世松[1] 吴亚楠 ZHU Shi-song;WU Ya-nan(School of Computer Science and Technology,Henan Polytechnic University,Jiaozuo 454000,China)

机构地区：[1]河南理工大学计算机科学与技术学院

出　　处：《仪表技术与传感器》2019年第10期122-126,共5页Instrument Technique and Sensor

基　　金：河南省国际科技合作项目(084300510065);河南省教育厅科学技术研究重点项目(13A520340);河南理工大学博士基金项目(B2010-95)

摘　　要：为有效滤除图像中椒盐噪声,同时能保留更多边缘和细节信息,构造了脉冲耦合神经网络(PCNN)与非对称裁剪中值滤波(UTMF)结合的去噪方法。设计自适应连接系数的改进型PCNN,提高对噪声的辨识;提出椒盐噪声在PCNN点火时间矩阵中的值具有极限性以定位噪声;在基于点火时间矩阵决策下对噪声图像进行UTMF处理。实验表明,该方法信噪比高、细节信息保护好、视觉效果佳、适应性强。Aiming to filter salt and pepper noise effectively and retain more edges and details in the image processing,a de-noising method was constructed by combining the Pulse Coupled Neural Network(PCNN)and the Unsymmetric Trimmed Median Filter(UTMF).An improved PCNN with adaptive connection coefficient was designed to improve the identification of noise pixel,and it proposed that the value of salt and pepper noise in PCNN ignition time matrix had the limit to locate noise pixel,then the noise image was processed by UTMF based on the ignition time matrix decision.Experiments shows that the method has higher signal-to-noise ratio,better detail information protection,better visual effect and more adaptive.

关 键 词：PCNN 自适应连接系数 点火时间矩阵 UTMF 椒盐噪声 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]