首页    期刊浏览 2024年07月08日 星期一
登录注册

文章基本信息

  • 标题:Multi-modal knowledge base generation from very high resolution satellite imagery for habitat mapping
  • 本地全文:下载
  • 作者:Ioannis Manakos ; Eleanna Technitou ; Zisis Petrou
  • 期刊名称:European Journal of Remote Sensing
  • 电子版ISSN:2279-7254
  • 出版年度:2016
  • 卷号:49
  • 期号:1
  • 页码:1033-1060
  • DOI:10.5721/EuJRS20164953
  • 摘要:Monitoring of ecosystems entails the evaluation of contributing factors by the expert ecologist. The aim of this study is to examine to what extent the quantitative variables, calculated solely by the spectral and textural information of the space-borne image, may reproduce verified habitat maps. 555 spectral and texture attributes are extracted and calculated from the image. Results reached an overall accuracy of 65% per object, 76% per pixel, and 77% in reproducing the original objects with segmentation. Taking into consideration the large number of different habitats queried and the lack of any ancillary information the results suggest the discriminatory power of the finally selected attributes. Potential and limitations are discussed.
  • 关键词:Texture analysis ; segmentation ; EUNIS habitats ; feature selection ; random forest ; ecosystem monitoring
国家哲学社会科学文献中心版权所有