地球信息科学学报 ›› 2021, Vol. 23 ›› Issue (7): 1272-1285.doi: 10.12082/dqxxkx.2021.200652

• 遥感科学与应用技术 • 上一篇    下一篇

融合面向对象和分水岭算法的山地湖泊提取方法

李文萍1,2(), 王伟1,*(), 高星1, 伍宇明1, 王学成1,2, 刘青1,2   

  1. 1.中国科学院地理科学与资源研究所 资源与环境信息系统国家重点实验室,北京 100101
    2.中国科学院大学,北京 100049
  • 收稿日期:2020-11-01 修回日期:2020-12-28 出版日期:2021-07-25 发布日期:2021-09-25
  • 通讯作者: 王伟
  • 作者简介:李文萍(1996— ),女,安徽淮南人,硕士生,主要从事遥感与GIS应用研究。E-mail: liwp.19s@igsnrr.ac.cn
  • 基金资助:
    科技部重点研发计划项目(YS2018YFGH000001);中国科学院战略性先导科技专项A类(XDA23090503);国家自然科学基金项目(41421001)

A Lake Extraction Method in Mountainous Regions based on the Integration of Object-Oriented Approach and Watershed Algorithm

LI Wenping1,2(), WANG Wei1,*(), GAO Xing1, WU Yuming1, WANG Xuecheng1,2, LIU Qing1,2   

  1. 1. Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, State Key Laboratory of Resources and Environmental Information System, Beijing 100101, China
    2. University of Chinese Academy of Sciences, Beijing 100049, China
  • Received:2020-11-01 Revised:2020-12-28 Online:2021-07-25 Published:2021-09-25
  • Contact: WANG Wei
  • Supported by:
    National Key Research and Development Plan of China(YS2018YFGH000001);Strategic Priority Research Program (Class A) of the Chinese Academyof Sciences(XDA23090503);National Natural Science Foundation of China(41421001)

摘要:

面向对象的方法提取湖泊,常常面临边界识别不精确的问题。本研究在面向对象方法的基础上,利用分水岭算法,解决湖泊边界识别问题。该方法初步将遥感影像划分为确定湖泊区域、潜在湖泊区域和背景;然后通过分水岭算法对潜在湖泊区域进行二次提取。研究选择昆仑-喀喇和喜马拉雅山脉区域的3个山地湖泊发育良好的区域作为实验区,利用Landsat系列影像验证该算法。实验结果表明该算法的用户精度、生产者精度和总体精度分别高达99.59%、98.47%和96.53%。相比于单一的面向对象方法,本文方法更适合于山地湖泊提取,能够更加准确地描绘湖泊的实际边界,也能够减弱面向对象方法中分割尺度和分类阈值对提取结果的影响。

关键词: Landsat, 湖泊, 山地, 面向对象, 影像分割, 分水岭算法, 水体指数, 遥感

Abstract:

The object-oriented methods for lake extraction from remote sensing images often have the problem of inaccurately identifying lake boundaries. This paper proposes a method to solve this problem by integrating the object-oriented approach with watershed algorithm. First, this method segments the target image into lakes, potential lake zone, and unknown region. Then, the unknown region will be refined using watershed algorithm. This work selected three mountainous regions with abundant lakes in Kunlun-Kara and Himalayas as the study area and used the Landsat images to evaluate the proposed method. The results show that the user's accuracy, producer's accuracy, and overall accuracy were up to 99.59%, 98.47%, and 96.53% respectively. Compared with single object-oriented method, the proposed method was more suitable for lake extraction in mountainous regions. Meanwhile, this method can not only accurately delineate the actual boundary of lakes, but also reduce the effect of segmentation scale and classification threshold on lake extraction result.

Key words: Landsat, lake, mountainous region, object-oriented, image segmentation, watershed algorithm, water index, remote sensing