一种新的PCNN实现方法及其在图像处理中的应用  

PCNN IMPLEMENTATION AND APPLICATIONS IN IMAGE PROCESSING

作　　者：方勇[1] 戚飞虎[1] 裴炳镇[1] 

机构地区：[1]上海交通大学计算机科学与工程系,上海200030

出　　处：《红外与毫米波学报》2005年第4期291-295,共5页Journal of Infrared and Millimeter Waves

基　　金：国家自然科学基金资助项目(60072029;60271033)

摘　　要：讨论了PCNN的实现方法及其在图像处理中的应用等问题.在PCNN的具体实现中,网络被强制为单循环网络,通过简化模型、阈值查找表、整数运算等技巧,降低了PCNN的时间复杂度.引入了时间索引图、指纹时间序列等概念,将PCNN与传统的图像处理技术结合起来,为实现PCNN的自动图像处理提供了一条新的途径;应用时间索引图,可以实现图像增强、边缘检测、图像分割和特征提取等.推导了PCNN的适用条件,并提出了一种实用的连接系数估计方法,对PCNN参数设置有着直接的指导作用.The PCNN implementation and applications in image processing were discussed, For the specific implementation of PCNNs, a PCNN was forced to be a single pass network. And simplified model, look-up threshold table, and integer operation were used to decrease the time complexity. The time-index image and fingerprint time series were introduced into PCNNs, and a PCNN was combined with the traditional techniques of image processing for automatic image processing, e.g. image enhancement, edge detection, image segmentation and feature extraction, by the time-index image. The applied conditions are derived for perfect image processing, and a practical estimation of the linking coefficient is proposed, which helps to determine PCNN parameters directly.

关 键 词：脉冲耦合神经网络 图像处理 特征提取 时间索引图 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]