Physics-Aligned Spectral Mamba: Decoupling Semantics and Dynamics for Few-Shot Hyperspectral Target Detection

2026-04-07

AI Summary

This work addresses the inefficiency and overfitting of full-parameter fine-tuning in few-shot hyperspectral target detection, as well as the neglect of spectral frequency structure and band continuity. To this end, the authors propose SpecMamba, a novel framework that integrates a frequency-aware lightweight adapter into a frozen Transformer backbone, decoupling semantic representation from spectral dynamics modeling. The method innovatively combines Discrete Cosine Transform (DCT) with the Mamba state space model to form the DCTMA module, which explicitly captures global spectral dependencies. Additionally, a Prior-Guided Triple Encoder (PGTE) and a Self-Supervised Pseudo-Label Mapping (SSPLM) mechanism are introduced to mitigate prototype drift and refine decision boundaries during inference. Experiments demonstrate that SpecMamba significantly outperforms existing approaches across multiple public datasets, achieving state-of-the-art performance in both detection accuracy and cross-domain generalization.

Abstract

Meta-learning facilitates few-shot hyperspectral target detection (HTD), but adapting deep backbones remains challenging. Full-parameter fine-tuning is inefficient and prone to overfitting, and existing methods largely ignore the frequency-domain structure and spectral band continuity of hyperspectral data, limiting spectral adaptation and cross-domain generalization.To address these challenges, we propose SpecMamba, a parameter-efficient and frequency-aware framework that decouples stable semantic representation from agile spectral adaptation. Specifically, we introduce a Discrete Cosine Transform Mamba Adapter (DCTMA) on top of frozen Transformer representations. By projecting spectral features into the frequency domain via DCT and leveraging Mamba's linear-complexity state-space recursion, DCTMA explicitly captures global spectral dependencies and band continuity while avoiding the redundancy of full fine-tuning. Furthermore, to address prototype drift caused by limited sample sizes, we design a Prior-Guided Tri-Encoder (PGTE) that allows laboratory spectral priors to guide the optimization of the learnable adapter without disrupting the stable semantic feature space. Finally, a Self-Supervised Pseudo-Label Mapping (SSPLM) strategy is developed for test-time adaptation, enabling efficient decision boundary refinement through uncertainty-aware sampling and dual-path consistency constraints. Extensive experiments on multiple public datasets demonstrate that SpecMamba consistently outperforms state-of-the-art methods in detection accuracy and cross-domain generalization.

Problem

• few-shot hyperspectral target detection
• spectral adaptation
• cross-domain generalization
• frequency-domain structure
• spectral band continuity

Innovation

• SpecMamba
• DCTMA
• few-shot hyperspectral target detection
• frequency-domain adaptation
• parameter-efficient learning