# SpectraMorph: Structured Latent Learning for Self-Supervised Hyperspectral Super-Resolution

arXiv: https://arxiv.org/abs/2510.20814 · HTML: https://mlsanity.com/paper/2510.20814/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 5 of its 12 references that are in the map.
- First posted on arXiv 23 Oct 2025.

## Builds on (most cited first, 5 of 5)

- [Fusformer: A Transformer-based Fusion Approach for Hyperspectral Image Super-resolution](https://mlsanity.com/paper/2109.02079/) (2021)
- [Model Inspired Autoencoder for Unsupervised Hyperspectral Image Super-Resolution](https://mlsanity.com/paper/2110.11591/) (2021)
- [Fourier-enhanced Implicit Neural Fusion Network for Multispectral and Hyperspectral Image Fusion](https://mlsanity.com/paper/2404.15174/) (2024)
- [A Spectral Diffusion Prior for Hyperspectral Image Super-Resolution](https://mlsanity.com/paper/2311.08955/) (2023)
- [SpectraLift: Physics-Guided Spectral-Inversion Network for Self-Supervised Hyperspectral Image Super-Resolution](https://mlsanity.com/paper/2507.13339/) (2025)

## Built on by (most cited first, 1 of 1)

- [HyperBench: Standardizing and Scaling Synthetic Evaluation for Hyperspectral Super-Resolution](https://mlsanity.com/paper/2605.21671/) (2026)

## Abstract (arXiv)

Hyperspectral sensors capture dense spectra per pixel but suffer from low spatial resolution, causing blurred boundaries and mixed-pixel effects. Co-registered companion sensors such as multispectral, RGB, or panchromatic cameras provide high-resolution spatial detail, motivating hyperspectral super-resolution through the fusion of hyperspectral and multispectral images (HSI-MSI). Existing deep learning based methods achieve strong performance but rely on opaque regressors that lack interpretability and often fail when the MSI has very few bands. We propose SpectraMorph, a physics-guided self-supervised fusion framework with a structured latent space. Instead of direct regression, SpectraMorph enforces an unmixing bottleneck: endmember signatures are extracted from the low-resolution HSI, and a compact multilayer perceptron predicts abundance-like maps from the MSI. Spectra are reconstructed by linear mixing, with training performed in a self-supervised manner via the MSI sensor's spectral response function. SpectraMorph produces interpretable intermediates, trains in under a minute, and remains robust even with a single-band (pan-chromatic) MSI. Experiments on synthetic and real-world datasets show SpectraMorph consistently outperforming state-of-the-art unsupervised/self-supervised baselines while remaining very competitive against supervised baselines.