A Cross-Hierarchical Difference Feature Fusion Network Based on Multiscale Encoder-Decoder for Hyperspectral Change Detection

TL;DR

This work tackles hyperspectral change detection by introducing CHDFFN, a cross-hierarchical, multiscale network that fuses bi-temporal differential features. It combines a multiscale feature extraction subnetwork with a dual-core channel-spatial attention mechanism, a spatial-spectral change feature learning module, and an adaptive fusion module to produce discriminative change representations. Extensive experiments on four public datasets demonstrate competitive gains in OA, KC, and F1 over state-of-the-art methods, validating the effectiveness of multiscale feature fusion and adaptive feature aggregation. The approach offers a path toward more robust HCD in complex, real-world scenarios and hints at future semi-supervised extensions to leverage unlabeled data.

Abstract

Hyperspectral change detection (HCD) is one of the core applications of remote sensing images, holding significant research value in fields like environmental monitoring and disaster assessment. However, existing methods often suffer from incomplete capture of multiscale spatial-spectral features and insufficient fusion of differential feature information. To address these challenges, this paper proposes a Cross-Hierarchical Differential Feature Fusion Network (CHDFFN) based on a multiscale encoder-decoder. Firstly, a multiscale feature extraction subnetwork is designed, taking the customized encoder-decoder as the backbone, combined with residual connections and the proposed dual-core channel-spatial attention module to achieve multi-level extraction and initial integration of spatial-spectral features. The encoder embeds convolutional blocks with different receptive field sizes to capture multiscale representations from shallow details to deep semantics. The decoder fuses the encoder's output via skip connections to gradually restore spatial resolution while suppressing background noise and redundancy. To enhance the model's ability to capture differential features between bi-temporal hyperspectral images, a spatial-spectral change feature learning module is designed to learn hierarchical change representations. Additionally, an adaptive high-level feature fusion module is proposed, dynamically balancing the contribution of hierarchical differential features by adaptively assigning weights, which effectively strengthens the model's capability to characterize complex change patterns. Finally, experiments on four public hyperspectral datasets show that compared with some state-of-the-art methods, the average maximum improvements of OA, KC, and F1 are 4.61%, 19.79%, and 18.90% respectively, verifying the model's effectiveness.

Figures (16)

• Figure 1: Overall structure diagram of CHDFFN.
• Figure 2: Structure of the multiscale feature extraction subnetwork.
• Figure 3: Structure of the residual block.
• Figure 4: Structure diagram of the multiscale encoder. From left to right: multiscale convolution module, axial attention.
• Figure 5: Structure of the decoder. From left to right: decoder, multi-head self-attention, and scaled dot-product attention.