基于卷积神经网络的遥感图像水体提取

Water extraction from remote sensing images based on CNN

为解决基于遥感图像监测地表水资源变化的问题,在深度学习的框架下,基于卷积神经网络(CNN)提出了用于遥感图像水体提取的模型。利用网络爬虫的方式,搜集遥感图像,并通过随机裁剪、数据清洗等方式构建训练、验证和测试数据集。通过对低层语义特征学习提取抽象的高层特征,基于提取的高层特征进行网络模型训练。实验结果表明:水体提取的精读准确率可高达96.28%,从而验证了所提模型对于遥感图像水体提取的可行性和有效性。

In order to solve the problem of monitoring changes of surface water resources based on remote sensing images,in the framework of deep learning,a model for water extraction from remote sensing images based on convolutional neural network(CNN)is proposed.WEB crawlers mode is used to collect remote sensing images,and construct training,verification,and test datasets through modes of random cropping,and data cleaning.High-level features are extracted by learning low-level semantic features,and network model training is performed based on the extracted high-level features.The experimental results show that the intensive reading accuracy rate of water body extraction can be as high as 96.28%,which verifies the feasibility and effectiveness of the proposed model for water body extraction from remote sensing images.