Content-Induced Spatial-Spectral Aggregation Network for Change Detection in Remote Sensing Images

Researchers have proposed two new networks for change detection in remote sensing images, one for general change detection and another for building damage assessment after earthquakes.

A team of researchers has introduced a content-guided spatial-spectral integration network (CSI-Net) for change detection in remote sensing images. CSI-Net integrates spatial and spectral information using a spatial reasoning module, a spectral difference module, and a content-guided integration module^[1]. The spatial reasoning module employs cascaded graph convolution blocks for global modeling, while the spectral difference module calculates the means and variances of features to reduce the impact of spectral differences in unchanged regions. This allows CSI-Net to learn changed features better and suppress spectral differences. In a separate study, researchers proposed a multi-scale interaction network (MSI-Net) for building change detection in earthquake scenarios, leveraging a newly created dataset called Turkey earthquake CD dataset (TUE-CD). Earthquakes are among the most destructive natural disasters, causing significant economic losses, and change detection can be applied to post-earthquake damage assessment using multi-temporal remote sensing images. However, current deep learning-based methods are limited due to the lack of datasets with short imaging intervals, a challenge addressed by TUE-CD. MSI-Net also tackles issues such as side-looking problems in post-event images^[2].