Current Issue Cover
用振子神经网络实现二值图象中的多目标分割

王磊1, 莫玉龙2, 戚飞虎1(1.上海交通大学计算机科学与工程系,上海 200030;2.上海大学通信与信息工程学院,上海 201800)

摘 要
利用基于视觉模型建立起来的局部激励、全局抑制振子神经网络(OSNN)实现二值图象中数字及字母的目标/背景分割;从理论上阐述了该网络实现目标/背景分割的机理 ,并分析了如何合理选择参数以利于成功分割。
关键词
Multiple Object Segmentation in Binary Images by Oscillator Neural Networks(OSNN)

()

Abstract
This paper proposes a method of segmenting numbers and characters from the background in the binary images based on locally excitatory and globally inhibitory oscillator neural network(OSNN) which was built according to the visual model. It also discusses the mechanism of object/background segmentation made by the neural net and the problem of how to select ing parameters resonablely in detail.
Keywords

订阅号|日报