FSCM: Frequency-Enhanced Spatial-Spectral Coupled Mamba for Infrared Hyperspectral Image Colorization

2026-05-13

AI Summary

This work addresses the challenges of false-color rendering for infrared hyperspectral imagery, which suffers from the absence of natural color and fine texture, often leading existing single-band coloring methods to introduce structural distortions and semantic ambiguities. To overcome these limitations, the authors propose a Frequency-enhanced Spatial-Spectral Coupled State Space Generative framework (FSCM), which uniquely integrates state space models with frequency-domain mechanisms. The core of FSCM is the Frequency-Spectral Block (FSB), combining wavelet decomposition, Fourier gating, and dual-stream hybrid gating to jointly model global spatial-spectral dependencies, recover high-frequency details, and enhance local structures. Additionally, an online semantic segmentation-guided loss is introduced to improve semantic consistency in complex scenes. Experimental results demonstrate that the proposed method significantly outperforms current approaches in both visual quality and semantic fidelity, thereby enhancing target recognition accuracy and human interpretability.

Abstract

Thermal infrared imaging is robust to illumination variations and smoke interference, making it important for all-weather perception. However, the lack of natural color and fine texture limits target recognition, human visual interpretation, and the transfer of visible-light models. Existing infrared colorization methods mainly rely on single-band images, where insufficient spectral cues may lead to structural distortion and semantic confusion. Although infrared hyperspectral images provide rich spectral responses and material information, existing single-band frameworks remain limited in modeling spatial-spectral coupling and weak texture details. To address these issues, this paper presents FSCM, a spectral-information-guided GAN framework. Within FSCM, a frequency-enhanced spatial-spectral state-space generator composed of cascaded FSB units is constructed. Each FSB integrates three complementary components: state-space modeling captures global spatial-spectral dependencies; the frequency enhancement module (FEM) combines multi-level wavelet decomposition and Fourier gating to recover structural contours, directional high-frequency details, and global frequency responses; and the dual-stream hybrid gating module (DGM) integrates deformation-aware sampling with sparse attention to enhance effective local structures and suppress background interference. Additionally, an online semantic segmentation-guided loss is introduced to constrain the generated results, improving semantic consistency in complex road scenes. Experiments show that FSCM outperforms existing infrared colorization methods in visual quality and semantic fidelity.

Problem

• infrared hyperspectral image colorization
• spatial-spectral coupling
• texture detail recovery
• semantic consistency
• structural distortion

Innovation

• Spatial-Spectral Coupling
• Frequency Enhancement
• State-Space Modeling
• Infrared Hyperspectral Colorization
• Semantic-Guided Generation