Multiscale diff-changed feature fusion network for hyperspectral image change detection.

Luo, Fulin; Zhou, Tianyuan; Liu, Jiamin; Guo, Tan; Gong, Xiuwen; Ren, Jinchang

Abstract

For hyperspectral images (HSI) change detection (CD), multi-scale features are usually used to construct the detection models. However, the existing studies only consider the multi-scale features containing changed and unchanged components, which is difficult to represent the subtle changes between bi-temporal HSIs in each scale. To address this problem, we propose a multi-scale diff-changed feature fusion network (MSDFFN) for HSI CD, which improves the ability of feature representation by learning the refined change components between bi-temporal HSIs under different scales. In this network, a temporal feature encoder-decoder sub-network, which combines a reduced inception module and a cross-layer attention module to highlight the significant features, is designed to extract the temporal features of HSIs. A bidirectional diff-changed feature representation module is proposed to learn the fine changed features of bi-temporal HSIs at various scales to enhance the discriminative performance of the subtle change. A multi-scale attention fusion module is developed to adaptively fuse the changed features of various scales. The proposed method can not only discover the subtle change of bi-temporal HSIs but also improve the discriminating power for HSI CD. Experimental results on three HSI datasets show that MSDFFN outperforms a few state-of-the-art methods.

Citation

LUO, F., ZHOU, T., LIU, J., GUO, T., GONG, X. and REN, J. 2023. Multiscale diff-changed feature fusion network for hyperspectral image change detection. IEEE transactions on geoscience and remote sensing [online], 61, article 5502713. Available from: https://doi.org/10.1109/TGRS.2023.3241097

• Journal Article Type: Article
• Acceptance Date: Jan 27, 2023
• Online Publication Date: Jan 31, 2023
• Publication Date: Dec 31, 2023
• Deposit Date: Feb 6, 2023
• Publicly Available Date: Feb 6, 2023
• Journal: IEEE Transactions on Geoscience and Remote Sensing
• Print ISSN: 0196-2892
• Electronic ISSN: 1558-0644
• Publisher: Institute of Electrical and Electronics Engineers (IEEE)
• Peer Reviewed: Peer Reviewed
• Volume: 61
• Article Number: 5502713
• DOI: https://doi.org/10.1109/tgrs.2023.3241097
• Keywords: Hyperspectral image; Change detection; Convolution encoder-decoder network; Multi-scale features; Attention fusion
• Public URL: https://rgu-repository.worktribe.com/output/1871941