点状地图符号的神经网络识别  

The recognition of map symbols with neural network

作　　者：何孝莹[1] 宋伟东[1] 韩颜顺[1] 

机构地区：[1]辽宁工程技术大学,辽宁阜新123000

出　　处：《测绘科学》2006年第1期125-127,共3页Science of Surveying and Mapping

基　　金：国家863计划(遥感数据处理软件)(2003AA135010)

摘　　要：基于神经网络的点状地图符号自动识别研究,设计了点状符号识别流程。采用四层BP网络模型,通过参数设置及输入输出设计优化网络,通过数据格式分析及内存调整优化程序,在程序实现过程中反复试验,总结出了参数的变化规律,实现了点状符号的简单、快速识别。The recognition of point map mark by using neural network is introduced in this paper together with the flowchart of the method. A fourlayer BP model is adopted and the network is improved by setting suitable input/output parameters. The software is optimized by analysis of data format and adjustment of memory. A rule of parameters changing is studied and valuable conclusions are drawn. The point map mark can be recognized easily with the new method.

关 键 词：点状符号 网络模型 参数调整 

分 类 号：P208[天文地球—地图制图学与地理信息工程]