Research on the Change Detection Method of Land Cover Based on Deep Learning
The research on automatic change detection of land cover is of great significance for the normalization of geographical national conditions monitoring.Combined with the full convolution neural network(FCN)and the twin(Siam)neural network,a full convolution twin network model(FCSCN)is designed.Through the construction of the sample database of urban land cover change,model training and testing,and precision evaluation,a deep learning model suitable for the detection of urban land cover change in Shenyang is obtained.By constructing a sample database of urban land cover change,then conducting model training and testing,and finally carrying out precision evaluation,a suit-able deep learning model for Shenyang urban land cover change detection is obtained.Taking the local area of the 2022 geographical national conditions monitoring project as the pilot,the practical exploration was carried out.The results showed that this method can improve work effi-ciency,and has certain practical reference for the high-frequency and full-coverage geographical national conditions monitoring.
deep learninggeographical national conditions monitoringland coverchange detectionconvolutional neural network
NETL
NSTL
万方数据
