Local Information-Enhanced Graph-Transformer for Hyperspectral Image Change Detection With Limited Training Samples

Hyperspectral image change detection (HSI-CD) is a challenging task that focuses on identifying the differences between multitemporal HSIs. The recent advancement of convolutional neural network (CNN) has made great progress on HSIs-CD. However, due to the limited receptive field, most CNN-based CD models trained with sufficient labeled samples cannot flexibly model the global information that is essential for distinguishing complex objects, thereby achieving relatively low performance. In this article, we propose a dual-branch local information-enhanced graph-transformer (D-LIEG) CD network to fully exploit the local-global spectral-spatial features of the multitemporal HSIs with limited training samples for change recognition. Specifically, the proposed network is composed of a cascaded of LIEG blocks, which jointly extracts local-global features by learning local information representation to enhance the information of graph-transformer. A novel graph-transformer is developed to model global spectral-spatial correlation between graph nodes, enabling the spectral information preservation of HSIs and accurate CD of areas with various sizes. Extensive experiments have proved that our method achieves significant performance improvement than other state-of-the-art methods on four commonly used HSI datasets.