欢迎访问《应用生态学报》官方网站,今天是 分享到:

应用生态学报 ›› 2005, Vol. 16 ›› Issue (2): 296-300.

• 研究论文 • 上一篇    下一篇

向海自然保护区土地覆盖分类研究

韩敏1, 程磊1, 唐晓亮1, 王教河2   

  1. 1. 大连理工大学电子与信息工程学院自动化系, 大连, 116023;
    2. 水利部松辽水利委员会, 长春, 130021
  • 收稿日期:2004-01-12 修回日期:2004-03-25 出版日期:2005-02-15 发布日期:2005-02-15
  • 通讯作者: 韩敏,女,1959年生,工学博士,教授,主要从事神经网络、专家系统、3S系统及混沌序引分析方面的研究,发表论文82篇.Tel:0411-84707847;E-mail:minhan@dlut.edu.cn.
  • 基金资助:
    国家自然科学基金重点资助项目(50139020).

Land cover classification in Xianghai Nature Reserve

HAN Min1, CHENG Lei1, TANG Xiaoliang1, WANG Jiaohe 2   

  1. 1. School of Electronic and Information Engineering, Dalian University of Technology, Dalian 116023, China;
    2. Songliao Water Resource Commission of Ministry of Water Resource, Changchun 130021, China
  • Received:2004-01-12 Revised:2004-03-25 Online:2005-02-15 Published:2005-02-15

摘要: 探讨了FuzzyARTMAP神经网络在向海自然保护区土地覆盖分类中的应用.全文阐述了FuzzyARTMAP网络结构及其采用的算法,提出一种引入遥感图像判读结果的警戒系数自动调整方法,能够解决人为选择警戒参数效率低、难以取得合适数值的问题.结果表明,具有警戒系数调整功能的FuzzyARTMAP神经网络能够有效的对向海自然保护区的TM影像进行分类,它与最大似然法和传统的FuzzyARTMAP神经网络相比对样本的依赖程度较低,分类精度较高.警戒系数自动调整方法根据遥感影像判读结果自动调整网络参数,提高了网络的运行速度.从分类结果可以看出,向海自然保护区目前开荒现象比较严重,草场有所退化,沼泽有荒漠化迹象,应当采取相应的措施保护向海湿地资源.

关键词: 遥感, 土地覆盖分类, Fuzzy ARTMAP, 向海自然保护区

Abstract: This paper discussed the application of Fuzzy ARTMAP neural network in land cover classification in Xianghai Nature Reserve.The structure and algorithm of Fuzzy ARTMAP neural network were described in detail,and an automatic adjustment method of vigilance parameter was put forward to solve the problem of searching the optimum value in the selection of vigilance parameter.The results showed that the automatic adjustment method could adjust network parameter automatically,and improve the running speed of network.In comparing with maximum likelihood classification method and traditional Fuzzy ARTMAP neural network,the Fuzzy ARTMAP neural network with the functioning of automatic adjustment depended less on samples and had higher accuracy,and thus,could effectively make the classification of TM image covering Xianghai Nature Reserve.It's shown from the classification that in the Xianghai Nature Reserve,farmland covered large area while grassland and marsh were facing degradation,and hence,corresponding countermeasures should be taken to improve the eco-environment of this Reserve.

Key words: Remote sensing, Land cover classification, Fuzzy ARTMAP, Xianghai Nature Reserve

中图分类号: