Implicit spatial-frequency fusion of hyperspectral and lidar data via kolmogorov-arnold networks

2026-05-13

AI Summary

This work addresses the challenges of hyperspectral image analysis in complex scenes, where spectral ambiguity, spatial heterogeneity, and strong coupling with geometric structures hinder effective modeling. Existing methods struggle to capture the nonlinear relationships and cross-domain synergy between hyperspectral and LiDAR data. To overcome these limitations, this study proposes IFGNet, which introduces Kolmogorov–Arnold Networks (KANs) into multimodal remote sensing fusion for the first time. Leveraging learnable spline-based activation functions, IFGNet adaptively models the highly nonlinear interactions between hyperspectral and LiDAR modalities. Furthermore, a LiDAR-guided implicit aggregation module enables geometry-aware fusion in both spatial and frequency domains. By transcending the constraints of fixed activations and linear weights in conventional CNNs or MLPs, IFGNet achieves state-of-the-art performance on the Houston 2013 and MUUFL datasets, consistently outperforming existing approaches in overall accuracy, average accuracy, and Cohen’s Kappa coefficient.

Abstract

Hyperspectral image (HSI) classification is challenging in complex scenes due to spectral ambiguity, spatial heterogeneity, and the strong coupling between material properties and geometric structures. Although LiDAR provides complementary elevation information, most HSI-LiDAR fusion methods rely on CNNs or MLPs with fixed activation functions and linear weights. These methods struggle to model structural discontinuities in LiDAR data, intricate spectral features of HSI, and their interactions. In addition, fusion of the two modalities in both spatial and frequency domains with LiDAR guidance remains underexplored. To address these issues, we propose the Implicit Frequency-Geometry Fusion Network (IFGNet), which leverages Kolmogorov-Arnold Networks (KANs) with learnable spline-based functions to adaptively capture highly nonlinear relationships between hyperspectral and LiDAR features. Furthermore, IFGNet introduces a LiDAR-guided implicit aggregation module in both spatial and frequency domains, enhancing geometry-aware spatial representations while capturing global structural patterns. Experiments on the Houston 2013 and MUUFL benchmarks demonstrate that IFGNet consistently outperforms existing fusion methods in overall accuracy, average accuracy, and Cohen's Kappa, while maintaining an efficient architecture.

Problem

• hyperspectral image classification
• LiDAR fusion
• spatial-frequency fusion
• spectral ambiguity
• structural discontinuities

Innovation

• Kolmogorov-Arnold Networks
• implicit fusion
• spatial-frequency domain
• LiDAR-guided aggregation
• hyperspectral-LiDAR fusion