Semisupervised SAR image change detection based on a siamese variational autoencoder

作者：

Highlights：

• This method introduces the variational autoencoder (VAE) for feature learning.

• A siamese VAE is designed to extract the spatial consistency information.

• Semi-supervised learning strategy is used to train the proposed model.

• Experimental results of our model are better than the state-of-the-art methods.

摘要

•This method introduces the variational autoencoder (VAE) for feature learning.•A siamese VAE is designed to extract the spatial consistency information.•Semi-supervised learning strategy is used to train the proposed model.•Experimental results of our model are better than the state-of-the-art methods.

论文关键词：Synthetic aperture radar (SAR) images,Change detection,Variational autoencoder,Siamese structure,Semisupervised learning

论文评审过程：Received 15 June 2021, Revised 19 August 2021, Accepted 20 August 2021, Available online 24 September 2021, Version of Record 24 September 2021.

论文官网地址：https://doi.org/10.1016/j.ipm.2021.102726