【作者】 卢志军；

【导师】 张军英；

【作者基本信息】 西安电子科技大学 ， 计算机应用， 2004， 硕士

【摘要】 本文在分析脉冲耦合神经网络(Pulse-coupled Neural Networks，以下简称PCNN)工作机理的基础上，着重研究了它在图像滤波和图像边缘检测中的应用。PCNN是近年来提出的一种新型网络，被称为第三代人工神经网络，它是通过模拟猫的大脑视觉皮层中同步脉冲发放行为而建立起来的一个简化模型。目前关于PCNN工作机理及其在图像处理、自动目标识别、注意、组合优化、人工生命等领域的研究和应用正得到国外广泛的重视。然而，国内在这方面的研究则可以说刚刚起步。本文的工作主要有：(1)通过对PCNN工作机理和运行行为的深入分析，揭示了PCNN的运行行为实际上是在对网络输入信息进行重新组织，并在此基础上给出了利用PCNN实现图像滤波(包括图像椒盐噪声和图像高斯噪声滤波)的算法。(2)结合图像边缘特性通过给网络加入侧抑制，提出了一个新的图像边缘检测算法。无论是滤波还是边缘检测，与其他方法相比，整个课题富有新意，而且在实际应用中表现出了明显的优越性，是PCNN在图像领域中的一个大胆尝试。更多还原

【Abstract】 This thesis firstly gives a detailed analysis of the mechanism of Pulse-coupled Neural Networks (Simplified as PCNN), and then emphasizes its applications on image noise removal and edge detection. PCNN is a new type of network and is called the third generation artificial neural network. It is a simplified model built through the simulation of the outbursts of synchronous pulses in the visual layer of a cat’s cerebra. At present, much attention is paid to the research on the mechanism of PCNN and its applications on image processing, automatic target recognition, combination and optimization, attention and artificial life abroad. Yet less work has been done at home. In this paper, the main work includes: (1) Through the detailed analysis of the mechanism of PCNN, we point out that the behaviors of PCNN actually means the reorganization of input information, and finally two methods (including method for salt & pepper noise and method for Gaussian noise) for image noise removal based on PCNN are presented. (2) Given the features of image edge, a new function named lateral inhibition is added to PCNN, and a new algorithm for image edge detection based on PCNN is also produced. The whole topic is full of new ideas, and shows its great advantages over other methods.更多还原