一种新的PCNN实现方法及其在图像处理中的应用

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．引入了时问索引图、指纹时问序列等概念，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．

PCNN IMPLEMENTATION AND APPLICATIONS IN IMAGE PROCESSING

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.