摘要
变化检测是实施跨比例尺空间数据更新的关键一环,也是难点问题。新旧数据间的差异信息不仅包括真实地理实体的变化信息,还包括地图综合引起的表达差异。空间数据更新过程中2种类型变化信息的甄别是一个复杂的决策过程,需要综合距离、面积、形状、方向等多种指标进行判断分析。鉴于上述问题,对线状水系目标进行研究,引入BP神经网络构建跨比例尺新旧数据变化识别模型,通过机器学习获得变化识别知识,从而解决多指标决策下的线状水系目标变化检测这一难点问题。利用真实数据开展实验分析,通过交叉验证方式发现5组实验大多数情形识别正确率高于90%,证明了所提方法的有效性。
Change detection is a key part of implementing cross-scale spatial data updates,and is also a difficult issue.The difference between the old and new data includes not only the change of real geographical entities,but also the difference in representation caused by map generalization.The screening of the two types of change information in the spatial data updating process is a complex decision-making process that requires a combination of distance,area,shape,direction and other indicators for judgement and analysis.Considering the above issues,this paper takes the update of linear river elements as the object,introduces BP neural network to build a model for identifying changes across scales of old and new data.The difficult problem of detecting changes in linear river targets under multi-indicator decision making is solved by gaining knowledge of change recognition through machine learning.Using real data for experimental analysis,it was found through cross validation that the recognition accuracy of most cases in the 5 experiments was higher than 90%,proving the effectiveness of the proposed method.
作者
吴超超
张崇善
姜楠
WU Chaochao;ZHANG Chongshan;JIANG Nan(Zhengyuan Geomatics Group Co.,Ltd.,Beijing 101300,China)
出处
《微型电脑应用》
2025年第3期200-204,共5页
Microcomputer Applications
