基于深度学习的遥感影像滑坡监测——以湖南涟源为例

Remote Sensing Image Landslide Monitoring Based on Deep Learning——Taking Lianyuan,Hunan as an Example

滑坡类地质灾害是湖南省主要的地质灾害。其中,涟源地区滑坡的成因、形态、规模与类别呈现多样性。文章通过多分辨率图像特征提取和深度学习,对提取的特征进行时序关联分析,结合滑坡先验知识,构建由模型与知识驱动的监测体系,提前预警滑坡灾害,以便做出应急举措,大幅度减少滑坡所带来的生命损失和财产损失。

Landslide type geological hazards are the main geological hazards in Hunan Province.The causes,morphology,scale and categories of landslides in Lianyuan area show diversity.The article uses multi resolution image feature extraction and deep learning to perform temporal correlation analysis on the extracted features,and combines landslide prior knowledge to construct a monitoring system driven by models and knowledge.It provides early warning of landslide disasters in order to take emergency measures and significantly reduce the loss of life and property caused by landslides.