Cross-layer and frequency fusion improve remote sensing image classification
Cross-Layer Refinement and Frequency-aware Fusion Network for Multi-Source Remote Sensing Image Classification
Computer Vision and Pattern Recognition
Summary
Classifying images from multiple remote sensing sources can be tricky because each type of image offers different kinds of useful information. The authors propose a new lightweight method called CRFNet that improves how these different images are combined and analyzed together. It uses special techniques to focus on important image details across multiple layers and to merge information in a way that highlights useful patterns. Tests on two real datasets show CRFNet works better than many existing methods.
What this means in practice
- •For geospatial analysts: Automatically classify land use and land cover by combining multiple remote sensing image types with enhanced accuracy.
- •For environmental monitoring teams: Improve detection of environmental changes by integrating spectral and spatial information from diverse satellite images.
Authors
Wenhan Xiao, Chuanzheng Gong, Feng Gao, Junyu Dong, Qian Du
Abstract
Multi-source remote sensing image classification benefits from complementary spectral and structural information. However, identifying discriminative cross-modal features remains challenging. This letter proposes a Cross-Layer Refinement and Frequency-aware Fusion Network (CRFNet) within a lightweight dual-branch architecture. Specifically, a Cross-Layer Refinement Attention (CLRA) module propagates selected attention relationships across successive layers and employs top relation selection to progressively strengthen discriminative spatial dependencies while suppressing irrelevant responses. Furthermore, a Frequency-aware Complementary Fusion (FCF) module combines frequency-wise interactions with cross-modal channel attention in the Fourier domain. The resulting representations are then converted into spatial-domain gates to selectively modulate complementary features. Experiments on the Houston 2013 and Augsburg datasets demonstrate that the proposed CRFNet consistently outperforms several state-of-the-art methods. The codes are publicly available at https://github.com/oucailab/CRFNet