Fast2Peak: High-Fidelity Restoration Algorithm by Single-Pixel Masking towards Real-Time Spectral Measurements
Abstract
Improving the temporal and spatial resolution of spectroscopy inevitably reduces the photons available to each measurement, pushing rapid measurements into photon-starved conditions where weak spectral features are obscured by stochastic background noise. Here, we present Fast2Peak (F2P), a self-supervised Transformer framework that transforms the spectral restoration task into a context-prediction problem. Given that physically constrained spectral signals exhibit strong local continuity, whereas the stochastic noise fluctuates randomly and does not possess the continuity across neighboring spectral pixels, single-pixel masking enables F2P to restore the predictable signal structure while suppressing unpredictable fluctuations even without clean reference spectra. Unlike conventional offline spectral restoration, F2P achieves high-fidelity restoration within milliseconds, enabling real-time "acquire-and-process" operation. We demonstrate this capability in time-resolved electrochemical SERS and hyperspectral Raman imaging, where spectral restoration is performed during data acquisition. In addition, F2P has demonstrated its robust generalization across different instruments and spectral modalities, including Raman, NMR, XRD and FTIR.
Related
- Phys-Mamba: Physics-informed selective state-space fusion network for high-fidelity underwater image restoration
- When Raman Spectral Fidelity and Bacterial Classification Accuracy Diverge
- New experimental measurements of the H2-CO2 Collision-Induced Absorption in the [3907-4770] cm−1 spectral range from 241 to 498 K
- Accelerating Proton Affinity Prediction with Multi-Fidelity Machine Learning
- Boric Acid-Assisted Thermal Restoration of an Air-Degraded O3-Type Sodium Layered Oxide Cathode
BibTeX
@misc{zhang2026fast2peak,
title = {{Fast2Peak: High-Fidelity Restoration Algorithm by Single-Pixel Masking towards Real-Time Spectral Measurements}},
author = {Nan-Nan Zhang and Si-Heng Luo and Xin-Di Pei and Wen-Xuan Li and Jun Wang and Yi-Fan Bao and Hao Ma and Guo-Kun Liu and Bin Ren},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.29753},
eprint = {2610.29753},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.29753},
}
RIS
TY - UNPB
TI - Fast2Peak: High-Fidelity Restoration Algorithm by Single-Pixel Masking towards Real-Time Spectral Measurements
AU - Nan-Nan Zhang
AU - Si-Heng Luo
AU - Xin-Di Pei
AU - Wen-Xuan Li
AU - Jun Wang
AU - Yi-Fan Bao
AU - Hao Ma
AU - Guo-Kun Liu
AU - Bin Ren
PY - 2026
DA - 2026/10/21
PB - arXiv
DO - 10.48550/arxiv.2610.29753
UR - https://arxiv.org/abs/2610.29753
AB - Improving the temporal and spatial resolution of spectroscopy inevitably reduces the photons available to each measurement, pushing rapid measurements into photon-starved conditions where weak spectral features are obscured by stochastic background noise. Here, we present Fast2Peak (F2P), a self-supervised Transformer framework that transforms the spectral restoration task into a context-prediction problem. Given that physically constrained spectral signals exhibit strong local continuity, whereas the stochastic noise fluctuates randomly and does not possess the continuity across neighboring spectral pixels, single-pixel masking enables F2P to restore the predictable signal structure while suppressing unpredictable fluctuations even without clean reference spectra. Unlike conventional offline spectral restoration, F2P achieves high-fidelity restoration within milliseconds, enabling real-time "acquire-and-process" operation. We demonstrate this capability in time-resolved electrochemical SERS and hyperspectral Raman imaging, where spectral restoration is performed during data acquisition. In addition, F2P has demonstrated its robust generalization across different instruments and spectral modalities, including Raman, NMR, XRD and FTIR.
ER -