Data-efficient linear hyperspectral imaging through risk-guided co-design
Abstract
Computational hyperspectral imaging reconstructs spatial-spectral information from limited optical measurements. Linear reconstructors are efficient and interpretable, but often underperform deep nonlinear models, which typically require substantial training data and computational resources. Here we present a theory-guided encoding-reconstruction co-design framework that achieves, to our knowledge, state-of-the-art linear reconstruction performance. We establish a theory for the exact five-term decomposition of reconstruction risk and introduce a method for diagnosing the statistical-coverage lower bound within a fixed principal component analysis envelope, which together inform the optimized design of hyperspectral imaging systems. The framework applies across coded-aperture snapshot spectral imaging (CASSI), diffractive optical element imaging and spectral filter array imaging, achieving 34.76 dB mean PSNR in single-shot CASSI, outperforming a deep baseline by 9.43 dB with only five source scenes, and matching empirical cross-domain risk within 0.2%. Our results establish a data-efficient and theoretically grounded route to high-performance linear hyperspectral reconstruction, while providing a quantitative basis for identifying whether reconstruction is limited by optical encoding, reconstruction design or statistical coverage.
Related
- Physics-Guided Multi-Channel Restoration for Efficient and Accurate Atomic Force Microscopy Imaging.
- Single-fiber hyperspectral imaging via nanophotonic disordered dispersion.
- Scan-less Hyperspectral Polarimetric Imaging Using Dual-Comb Single-Pixel Imaging
- Hyperspectral Unmixing using a Structured Linear Mixing Model with Additive Spectral Variability
- High Resolution Hyperspectral Quantum Infrared Imaging with Undetected Photons
BibTeX
@misc{hong2026dataefficient,
title = {{Data-efficient linear hyperspectral imaging through risk-guided co-design}},
author = {Shenzhan Hong and Shuxing Yang and Taojun Ying and Fujia Chen and Zongyin Yang and Hongsheng Chen and Yihao Yang},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.34935},
eprint = {2610.34935},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.34935},
}
RIS
TY - UNPB
TI - Data-efficient linear hyperspectral imaging through risk-guided co-design
AU - Shenzhan Hong
AU - Shuxing Yang
AU - Taojun Ying
AU - Fujia Chen
AU - Zongyin Yang
AU - Hongsheng Chen
AU - Yihao Yang
PY - 2026
DA - 2026/10/26
PB - arXiv
DO - 10.48550/arxiv.2610.34935
UR - https://arxiv.org/abs/2610.34935
AB - Computational hyperspectral imaging reconstructs spatial-spectral information from limited optical measurements. Linear reconstructors are efficient and interpretable, but often underperform deep nonlinear models, which typically require substantial training data and computational resources. Here we present a theory-guided encoding-reconstruction co-design framework that achieves, to our knowledge, state-of-the-art linear reconstruction performance. We establish a theory for the exact five-term decomposition of reconstruction risk and introduce a method for diagnosing the statistical-coverage lower bound within a fixed principal component analysis envelope, which together inform the optimized design of hyperspectral imaging systems. The framework applies across coded-aperture snapshot spectral imaging (CASSI), diffractive optical element imaging and spectral filter array imaging, achieving 34.76 dB mean PSNR in single-shot CASSI, outperforming a deep baseline by 9.43 dB with only five source scenes, and matching empirical cross-domain risk within 0.2%. Our results establish a data-efficient and theoretically grounded route to high-performance linear hyperspectral reconstruction, while providing a quantitative basis for identifying whether reconstruction is limited by optical encoding, reconstruction design or statistical coverage.
ER -