基于国产高分辨率卫星影像的城市水体自动提取研究  

Automatic Extraction of Urban Water Bodies Based on DomesticHigh Resolution Satellite Images

作　　者：王陆培 WANG Lupei(Shanghai Puhai Surveying and Mapping Co.,Ltd.,Shanghai 201399,China)

机构地区：[1]上海浦海测绘有限公司,上海201399

出　　处：《城市勘测》2024年第2期88-92,共5页Urban Geotechnical Investigation & Surveying

基　　金：国家自然科学基金面上项目(42171431)。

摘　　要：基于卫星影像的水体提取已经成为遥感应用的重要方向,传统基于遥感的水体自动提取在特征简单的水体上已取得较好效果,但对于城市复杂环境下的水体却存在不同程度的误提及漏提的问题,尤其是建筑物阴影及细小水体造成的影响。针对此问题,本研究选用国产高分辨率卫星(吉林一号)影像,选用同时具备分类能力及像素定位能力的深度学习模型,制作城市水体样本并迭代训练,实现嘉定城区范围水体的自动快速精确提取,在一定程度上解决了城市水体的提取难题。Water extraction based on satellite images has become an important direction of remote sensing applications,and the traditional automatic extraction of water bodies based on remote sensing has achieved good results in water bodies with simple characteristics,but there are different degrees of misreferences and omissions for water bodies in complex urban environments,especially the impact of building shadows and small water bodies.In order to solve this problem,this study uses domestic high-resolution satellite(Jilin No.1)images,and selects a deep learning model with both classification and pixel positioning capabilities,and makes urban water samples and iterative training,so as to realize the automatic rapid and accurate extraction of water bodies in Jiading urban area,which solves the extraction problem of urban water bodies to a certain extent.

关 键 词：城市水体 国产高分辨率影像 深度学习 U-Net 

分 类 号：TP753[自动化与计算机技术—检测技术与自动化装置]