Multilevel feature aggregation and enhancement network for remote sensing change detection

Abstract

Remote sensing change detection refers to the process of identifying and extracting changes in objects within the same geographical region over multiple periods. With the increasing spatial resolution of remote sensing images, the detection of minor changes has become a challenging task. We introduce a multilevel feature aggregation and enhancement network to tackle this issue. Specifically, we propose a multilevel feature aggregation module to aggregate the distinct features extracted from each image, which strengthens the feature representation capability. Subsequently, a difference parallel mapping module is designed to perceive information at different scales by refining the fused features. In addition, our guided change enhancement module captures local and long-range dependencies in multilevel features, improving the network's accuracy in identifying changing regions. Based on a basic shared weight Siamese backbone without complex structures, our model outperforms other state-of-the-art methods on three datasets in terms of both efficiency and effectiveness.

Guided filter-based self-supervised remote sensing image change detection with global contrast

“…[6][7][8][9][10][11] Among these deep learning methods, supervised learning methods have performed successfully in remote sensing image CD due to their powerful modeling and learning capabilities. [12][13][14][15][16] For example, Deng et al 15 incorporated depthwise separable convolution and multi-headed selfattention in UNet++, which effectively enhances the network's ability to extract local and global information. Yan et al 16 proposed a multilevel feature aggregation and enhancement network to sense information at different scales and capture their dependencies.…”

“…[12][13][14][15][16] For example, Deng et al 15 incorporated depthwise separable convolution and multi-headed selfattention in UNet++, which effectively enhances the network's ability to extract local and global information. Yan et al 16 proposed a multilevel feature aggregation and enhancement network to sense information at different scales and capture their dependencies. However, all supervised methods in deep learning require a large amount of training data with precise labeling to train, which requires a lot of resources for manual labeling, which is extremely difficult in the field of remote sensing.…”