改进型脉冲耦合神经网络在图像处理中的动态行为分析  

Analysis of Dynamic Behaviors of Improved Pulse Coupled Neural Network in Image Processing

作　　者：姚畅[1] 陈后金[1] 李居朋[1] 

机构地区：[1]北京交通大学电子信息工程学院,北京100044

出　　处：《自动化学报》2008年第10期1291-1297,共7页Acta Automatica Sinica

基　　金：新世纪优秀人才支持计划(50051);教育部科技重点项目(106030)资助~~

摘　　要：为了进一步拓展脉冲耦合神经网络(Pulse coupled neural network,PCNN)在图像处理领域的应用深度,本文对改进型PCNN在图像处理中的动态行为结合网络参数进行了具体的理论分析和公式推导,明确了PCNN的动态行为特性受网络参数变化的影响,并给出了保证其所有神经元只点火一次的网络参数确定准则,最后将改进型PCNN应用于实际图像分割中.实验结果证明了其动态行为特性与参数确定准则之间关系的正确性和算法的有效性.In order to further exploit the application depth of pulse coupled neural network （PCNN） in image processing, combined with network parameters, dynamic behaviors of an improved PCNN are analyzed theoretically and deduced by formula in detail. The influence of parameters＇ change on dynamic behaviors is proved, and the rule of parameters determination is proposed under the condition guaranteeing each neuron only fires once. Finally, the experiments implemented on the practical image segmentation indicate that the relationship between dynamic behaviors and parameters determination is accurate and the algorithm is effective.

关 键 词：脉冲耦合神经网络 动态行为 图像处理 神经元 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]