复杂背景下水体提取方法适用性研究  

Study on the applicability of water extraction methods under complex background

作　　者：高广旭 武永斌 卢小平[1] 周俊利 GAO Guangxu;WU Yongbin;LU Xiaoping;ZHOU Junli(Key Laboratory of Mine Space Information Technology, Ministry of Natural Resources, Henan Polytechnic University, Jiaozuo Henan 454003, China;Henan Institute of Remote Sensing Surveying and Mapping, Zhengzhou Henan 450003, China)

机构地区：[1]河南理工大学矿山空间信息技术自然资源部重点实验室,河南焦作454003 [2]河南省遥感测绘院,河南郑州450003

出　　处：《北京测绘》2022年第4期452-456,共5页Beijing Surveying and Mapping

摘　　要：研究复杂背景下最佳水体提取方法,已成为近年来水体提取研究的热门话题。本文以南湾水库为实验对象,基于光谱特性和分类的方法探究复杂环境下水体提取效果。研究表明:在各种方法中,以改进的归一化差异水体指数(modified normalized difference water index,MNDWI)以及结合该指数和归一化差异建筑指数(normalized difference building index,NDBI)的随机森林算法提取效果最好,多波段谱间关系法提取精度最低。复杂背景下高反射率建筑群及非均匀水质等因素均对水体提取造成干扰,可联合多种方法综合提取水体面积。The study of the best water extraction method under complex background has become a hot topic in water extraction research in recent years. Taking the Nanwan Reservoir as the object, based on the spectral characteristics and supervised classification method, the extraction effect of water bodies in complex environments was explored. The study showed that among various methods, the modified normalized differencewater index (MNDWI) and the random forest algorithm combining this index and the normalized difference building index (NDBI) had the best extraction effects, then theMulti-band spectral relationship method had the lowest extraction accuracy. The factors such as high-reflectivity buildings and non-uniform water quality in a complex background interfere with the extraction of water bodies, so a variety of methods can be combined to comprehensively extract the water body area.

关 键 词：南湾水库 水体提取 改进的归一化差异水体指数 复杂背景 

分 类 号：P237[天文地球—摄影测量与遥感]