自组织特征映射神经网络及其在彩色地图自动分层中的应用

详细信息

Self-organization Feature Mapping Artificial Neural Network and its Application in Map Segmentation

Dept. Of Photogrammetry and Remote Sensing, WTUSM, Luoyu Road 39, Wuhan, China, 430070

摘要: 地图数据分层是地图识别系统的一个重要组成部分。由于传统的基于Bayes准则识别方法要求已知概率分布密度,当实际概率分布与假设条件存在差异时,无法取得满意的识别效果。神经网络方法可以克服这些缺点。本文利用Kohonen提出的自组织特征映射网络的学习、识别能力,实现了对彩色地图扫描数据的自动分层。
- 神经网络 /
- 侧抑制 /
- 自组织特征映射 /
- 彩色地图数据分层
Abstract: The map segmentation is an important subsystem in automatic map recognition system.This paper discussed the ability of adaptive learning in self-orgnization artificial neural network and used it in map segmentation.The results of segmentation illustrated the efficiency of unsupervised learning and classification in self-orgnization artificial neuralwork.
- artificial neuralwork /
- lateral inhibition /
- self-orgnizing feature map /
- color map segmentation