论文部分内容阅读
本文提出了一种利用神经网络实现灰度形态滤波的方法。文中首先简述了形态学的基本理论，然后给出了灰度形态学理论的两种最基本运算（膨胀和腐蚀）的神经网络实现方法。网络中的连接权值为形态滤波的结构元素，按照δ学习规则，自适应地对结构元素进行学习训练。该方法计算简单，速度快，对于提高灰度形态滤波的性能有显著成效。
In this paper, a method of using neural network to realize gray morphological filtering is proposed. In this paper, the basic theory of morphology is briefly introduced. Then, two basic methods of gray network theory (expansion and erosion) neural network implementation are given. The connection weights in the network are the structural elements of morphological filtering, and the structural elements are adaptively trained according to δ learning rules. The method is simple, fast and has a significant effect on the performance of gray-scale morphological filtering.