华东师范大学学报(自然科学版) ›› 2010, Vol. 2010 ›› Issue (4): 26-34.

• 环境工程 地理学 • 上一篇    下一篇

结合GIS数据的神经网络湿地遥感分类方法:以上海崇明岛东滩湿地为例

栗小东1， 过仲阳1， 朱燕玲1， 戴晓燕2   

  1. 1. 华东师范大学 地理信息科学教育部重点实验室，上海 200062； 2. 华东师范大学 河口海岸学国家重点实验室，上海 200062
  • 收稿日期:2009-06-01 修回日期:2009-10-01 出版日期:2010-07-25 发布日期:2010-07-25
  • 通讯作者: 过仲阳

Artificial neural network classification of wetland integrating GIS data: A case study of Dongtan wetland in Chongming, Shanghai

LI Xiao-dong1, GUO Zhong-yang1, ZHU Yan-ling1, DAI Xiao-yan2   

  1. 1. Key Laboratory of Geographic Information Science, Ministry of Education, East China Normal University, Shanghai 200062, China; 2. State Key Laboratory of Estuarine and Coastal Research, East China Normal University, Shanghai 200062, China
  • Received:2009-06-01 Revised:2009-10-01 Online:2010-07-25 Published:2010-07-25
  • Contact: GUO Zhong-yang

PDF

1775

摘要： 以上海崇明岛东滩湿地为例，利用改进的BP算法结合主成分分析，将光谱信息的主成分、NDVI、MNDWI以及GIS数据作为神经网络的输入参数对东滩湿地进行神经网络分类.结果表明，神经网络分类能够有效的提高分类的精度，适合湿地分类.

关键词: 湿地, 遥感分类, 神经网络, 湿地, 遥感分类, 神经网络

Abstract: This paper took Dongtan wetland in Chongming Island, Shanghai, as a case study; using the PCA outputs of TM surface feature spectrum, NDVI, MNDWI, DEM and the GIS data as inputting elements of an Artificial Neural Network (ANN), combined with improved BP algorithm, an ANN classification was applied to the Dongtan wetland. The results show that the ANN classification method improves the classification accuracy, and can effectively distinguish those objects with similar TM spetra.

Key words: remote sensing classification, ANN, wetland, remote sensing classification, ANN