测绘通报 ›› 2019, Vol. 0 ›› Issue (8): 68-71,77.doi: 10.13474/j.cnki.11-2246.2019.0254

• 学术研究 • 上一篇    下一篇

非监督分类的冬小麦种植信息提取模型

王冬利1,2, 张安兵1,2, 赵安周1,2, 李静1,2   

  1. 1. 河北工程大学矿业与测绘工程学院, 河北 邯郸 056038;
    2. 河北工程大学河北省煤炭资源综合开发与利用协同创新中心, 河北 邯郸 056038
  • 收稿日期:2019-03-17 修回日期:2019-05-13 出版日期:2019-08-25 发布日期:2019-09-06
  • 通讯作者: 张安兵。E-mail:zhanganbing@hebeu.edu.cn E-mail:zhanganbing@hebeu.edu.cn
  • 作者简介:王冬利(1979-),男,博士,讲师,主要从事遥感在资源、环境和农业领域的应用研究。E-mail:wangdongli@hebeu.edu.cn
  • 基金资助:
    国家863计划子课题(2015AA123901);河北省自然科学基金(D2017402159);河北省高等学校科学技术研究重点项目(ZD2018230);河北省高等学校科学技术研究青年拔尖人才项目(BJ2018043)

Extraction model of winter wheat planting information based on unsupervised classification

WANG Dongli1,2, ZHANG Anbing1,2, ZHAO Anzhou1,2, LI Jing1,2   

  1. 1. School of Mining and Geomatics, Hebei University of Engineering, Handan 056038, China;
    2. Collaborative Innovation Center for the Comprehensive Development & Utilization of Coal Resources in Hebei Province, Hebei University of Engineering, Handan 056038, China
  • Received:2019-03-17 Revised:2019-05-13 Online:2019-08-25 Published:2019-09-06

摘要: 为了解决在区域冬小麦种植信息遥感提取过程中监督学习算法存在的需要地面样本数据支持、流程复杂、人为干扰因素多及自动化程度低等问题,本文以非监督分类为核心,结合多尺度技术,提出了一种新的非监督分类冬小麦种植信息提取模型。选取河北省辛集市为典型试验区,以2014年高分一号数据为数据源,对本文提出的模型进行实例验证。试验结果表明:该模型的Kappa系数为0.88,整体精度为94.00%;对于研究区内的冬小麦,在无需训练样本、人为干扰因素少等条件下,该模型具有与监督分类相似的提取精度,能够满足冬小麦种植信息地面遥感监测的需求。

关键词: 非监督分类, 冬小麦, 植被指数, 高分一号, 多尺度

Abstract: There are some problems of supervised learning algorithm in remote sensing extraction of regional winter wheat planting information, such as heavy dependence on ground sample data, complex process, too many artificial interference factors and low degree of automation, etc. In order to solve those problems, this paper proposed a model of winter wheat extraction which took unsupervised classification technology as the core and combined with multi-scale technology. It verified the accuracy and validity of the model proposed in this paper that took Xinji city of Hebei province as a typical experimental area and used the GF-1 remote sensing data in 2014. The experimental results show that the overall accuracy of the model is 94.00%, and the Kappa coefficients is 0.88. For winter wheat in the study area, the model can achieve the supervised classification extraction accuracy without training sample data and less artificial interference factors. So, the model can meet the requirements of ground remote sensing monitoring for winter wheat planting information.

Key words: unsupervised classification, winter wheat, vegetation index, GF-1, multi-scale

中图分类号: