基于深度学习的湖泊水体提取方法研究  

Research on lake water extraction method based on deep learning

作　　者：严传勇 Yan Chuanyong(Huizhou City Huayu Water Resources and Hydropower Engineering Survey and Design Co.,Ltd.,Huizhou516000,China)

机构地区：[1]惠州市华禹水利水电工程勘测设计有限公司,广东惠州516000

出　　处：《吉林水利》2024年第12期50-54,共5页Jilin Water Resources

摘　　要：湖泊是全球水资源的重要组成部分,湖泊水体提取关系到水文过程的研究、水资源量的评估等。遥感技术在探测范围、探测时效方面具有优势,被广泛应用于湖泊水体提取中,而机器学习的发展使基于遥感影像的水体提取技术获得了更大发展。本研究基于深度学习方法,通过提高水体提取精度,减少网络训练时间,优化边缘精度,对提取方法进行改进,并验证方法的准确性。结果表明,湖泊水体提取尤其是边缘精度得到有效提升,可为相关研究提供借鉴。Lakes are an important part of global water resources.The extraction of lake water surface is related to the study of hydrological processes and the assessment of water resources.Remote sensing technology is widely used in surface extraction due to its advantages in detection range and time.The development of machine learning has made great progress in water surface extraction based on remote sensing images.Using deep learning method,this research improves the extraction method by improving the extraction accuracy,reducing the network training time,and optimizing edge accuracy,and verifies the effectiveness of the method.The results show that lake surface extraction,especially edge accuracy,has been effectively improved,which can provide reference for related research.

关 键 词：水体提取 深度学习 遥感 

分 类 号：P237[天文地球—摄影测量与遥感]