欢迎访问《中国农学通报》，
中国农学通报

中国农学通报 ›› 2019, Vol. 35 ›› Issue (10): 137-141.doi: 10.11924/j.issn.1000-6850.casb18040059

所属专题： 耕地保护

• 农业科技信息 • 上一篇    下一篇

基于GF-1卫星数据与面向对象分类的达里诺尔湿地自然保护区土地覆盖信息提取

王艳琦1,2, 秦福莹3, 银山3, 彭秀清3   

  1. 1.内蒙古自治区呼和浩特市赛罕区昭乌达路81号内蒙古师范大学地理科学学院;2.010022;3.内蒙古师范大学地理科学学院
  • 收稿日期:2018-04-13 修回日期:2019-03-14 接受日期:2018-10-31 出版日期:2019-04-03 发布日期:2019-04-03
  • 通讯作者: 秦福莹
  • 基金资助:
    内蒙古师范大学研究生科研创新基金资助项目“基于GF-1 遥感影像对达里诺尔保护区土地覆盖分类研究”(CXJJS17096)；内蒙古自治区 研究生教育创新计划资助项目“基于GF-1 遥感影像对达里诺尔保护区土地覆盖分类研究”(S20171013508)；国家自然科学基金项目“基于高分光学 与高分极化SAR 的蒙古高原湿地退化指示种识别研究”(61661045)；科技基础资源调查项目“蒙古国基础地理要素与土地利用/覆被调查” (2017FY101301-4)；内蒙古科技财政项目“畜牧业生产服务云平台”(KCBJ2018007)；内蒙古自治区科技厅项目“中蒙俄跨境灾害遥感监测与风险评 估”(2018-ZME-KJXT-14)。

Land Cover Information Extraction of Dalinuoer Wetland Nature Reserve Based on GF-1 Satellite Data and Object-oriented Classification

  • Received:2018-04-13 Revised:2019-03-14 Accepted:2018-10-31 Online:2019-04-03 Published:2019-04-03

PDF (PC)

274

摘要： 以达里诺尔湿地自然保护区为研究区，基于国产GF-1遥感影像，采用面向对象和传统目视解译的分类方法对研究区土地覆盖遥感信息进行提取，并对其结果进行对比分析，采取混淆矩阵对面向对象分类结果进行精度验证。结果表明：（1）充分利用了GF-1遥感影像的光谱信息，面向对象分类采取试错法确定最优分割尺度为550，形状和紧致度因子分别为0.6和0.5，各波段权重均为1。（2）面向对象分类总体分类精度达98.22％，KAPPA系数为0.96；（3）面向对象分类方法可快速准确提取类型较为复杂的土地覆盖信息，为内陆湿地精准快速提取研究区土地覆盖分类信息提供参考，以期为湿地遥感业务化监测提供技术规范。

关键词: 稳定氮肥, 稳定氮肥, 油菜, 产量, 施用方式, 氮肥效率

Abstract: Based on domestic GF-1 remote sensing imagery, taking the Darinor Wetland Nature Reserve as the research area, we extracted land cover remote sensing information by adopting object-oriented and traditional visual interpretation classification methods, and compared and analyzed the results, verified the accuracy of the object-oriented classification results by using the confusion matrix. The results showed that: (1) spectral information of GF-1 remote sensing images was fully utilized, and the object-oriented classification method with trial-and-error to determine the optimal segmentation scale of 550, and the shape and compactness factors were 0.6 and 0.5, and each band weight was 1; (2) the overall classification accuracy of object-oriented classification was up to 98.22%, and the KAPPA coefficient was about 0.96; (3) land cover information of relatively complex types was extracted by the object-oriented classification method quickly and accurately, which provided a reference for accurate and rapid extraction of land cover for the inland wetland to provide technical specifications for wetland remote sensing operational monitoring.