自组织人工神经网络在立地条件遥感分类研究中的应用  

APPLICATION OF SELF-ORGANIZED ARTIFICIAL NERVE NETWORK IN SITE CLASSIFICATION BY REMOTE SENSING

作　　者：李元科[1] 全志杰[1] 张美亮 

机构地区：[1]西北林学院,陕西杨凌712100

出　　处：《陕西林业科技》1998年第4期24-27,33,共5页Shaanxi Forest Science and Technology

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.

关 键 词：神经网络 立地条件类型 遥感 地理信息系统 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]