摘要
运用以航空遥感图像为主的信息源,获取立地要素信息,在ARC/INFO系统支持下,建立空间信息库,选取550个样本数据,训练自组织人工神经网络,然后对159个“未知”样本进行立地分类预测和容错检验。结果表明,该模型的分类、容错能力强,综合了遥感图像目视判设与计算机自动分类的优点,开拓了遥感与GIS技术相结合进行智能化立地条件分类研究的新途径。
By using ARC/INFO system, spatial information storehouse on site condition was constructed from aerial remote sensing photo, 550 specimen was choose to train self-organized artificial nerve network, and 159 unkown sample was used to forecast the site classification and to test the ability of tolerane error.The results showed that the model possessed a powerful ability in site classification and in tolerance error,and also well synthesized the advantages of remote sensing photo interpretation by eyes and computer automatic recognition. The model opened up a new way for the intellectual site classification with the techniques combinated remote sensing and GIS.
出处
《陕西林业科技》
北大核心
1998年第4期24-27,33,共5页
Shaanxi Forest Science and Technology
关键词
神经网络
立地条件类型
遥感
地理信息系统
Self-organized artificial nerve network
Site type
Remote sensing
Geographic information system(GIS)
Ningxia