摘要
在高分辨率遥感影像水体提取任务中,基础UNet网络提取精度较低,在规则水体的边缘提取上不够清晰,且对于细小水体提取不够完全。对此,文章提出了一种基于TransU-Net网络模型的水体提取方法。其中,TransU-Net网络模型能够完整有效地对水体边缘和局部特征进行提取。实验结果表明,该方法优于基础UNet方法,提取精度相较基础UNet提高了3.63%,能够有效完成高分辨率遥感影像水体提取任务。
In the task of water extraction from high-resolution remote sensing images,the accuracy of the basic UNet network is low,and the edge extraction of regular water bodies is not clear enough,and the extraction of small water bodies is not complete enough.In this regard,this article proposes a water extraction method based on the TransU-Net network model.Among them,the TransU-Net network model can fully and effectively extract edge and local features of water bodies.The experimental results show that this method is superior to the basic UNet method,with an extraction accuracy improvement of 3.63%compared to the basic UNet method,and can effectively complete the task of extracting water bodies from high-resolution remote sensing images.
作者
韩久春
袁啸宇
杜海峰
张玉明
邱征
HAN Jiuchun;YUAN Xiaoyu;DU Haifeng;ZHANG Yuming;QIU Zheng(Anhui Construction Engineering Traffic&Shipping Group Co.,Ltd.,Hefei 230011,China;College Civil Engineering,Hefei University of Technology,Hefei 230009,China)
出处
《计算机应用文摘》
2024年第10期45-47,共3页
Chinese Journal of Computer Application
