LIPS-PC: Learning Illumination Patterns for Single-Pixel Classification
Abstract
In single-pixel (ghost) imaging, known illumination patterns probe an object and a detector with no spatial resolution returns one number per pattern. Each measurement costs acquisition time and photons and, for ionizing probes, contributes to delivered dose. Motivated by automated radiographic defect screening, where a class label suffices, we introduce Learning Illumination Patterns for Single-Pixel Classification (LIPS-PC). The patterns are learned end-to-end while the measurement operator remains one fixed linear matrix, identical for every input. We parameterize it as a residual correction to a two-dimensional discrete cosine transform (2D-DCT), D=Dbase+Δ, and jointly train the correction and a compact classifier using cross-entropy, without a reconstruction objective. This yields a task-driven linear embedding through optimization. We evaluate LIPS-PC numerically on MNIST and Fashion-MNIST under a controlled protocol, then on RIAWELC, a real industrial X-ray weld-radiograph dataset. Across five budgets, M∈8,16,32,64,128, corresponding to sampling ratios from below 1% to 12.5%, the learned operator outperforms the best tested fixed transform by 9-15 percentage points at M=8 on all three datasets. Its eight-measurement accuracy requires roughly 16-32 measurements with the best fixed transform, indicating a 2× to 4× saving at comparable accuracy. The advantage generally shrinks as M increases, with a small reversal on RIAWELC at higher budgets. These results show that learning what to measure is especially valuable when measurements are scarce.
Related
- Coprime Vernier-based parallel single-pixel imaging for depth-constraint-free 3D imaging
- Robust single-pixel imaging for high-speed rotating objects with large rotational speed variations
- Post-fabrication trimming of a 1024-pixel, single-photon counting, Microwave Kinetic Inductance Detector array with a pixel pitch of 150 micron
- Single-event upset response of the ITkPixV1.1 pixel readout chip with 24 GeV proton irradiation
- Fast2Peak: High-Fidelity Restoration Algorithm by Single-Pixel Masking towards Real-Time Spectral Measurements
BibTeX
@misc{porat2026lipspc,
title = {{LIPS-PC: Learning Illumination Patterns for Single-Pixel Classification}},
author = {Tomer Porat and Talya Eden and Eliahu Cohen},
howpublished = {arXiv},
year = {2026},
month = nov,
doi = {10.48550/arxiv.2611.04780},
eprint = {2611.04780},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2611.04780},
}
RIS
TY - UNPB
TI - LIPS-PC: Learning Illumination Patterns for Single-Pixel Classification
AU - Tomer Porat
AU - Talya Eden
AU - Eliahu Cohen
PY - 2026
DA - 2026/11/03
PB - arXiv
DO - 10.48550/arxiv.2611.04780
UR - https://arxiv.org/abs/2611.04780
AB - In single-pixel (ghost) imaging, known illumination patterns probe an object and a detector with no spatial resolution returns one number per pattern. Each measurement costs acquisition time and photons and, for ionizing probes, contributes to delivered dose. Motivated by automated radiographic defect screening, where a class label suffices, we introduce Learning Illumination Patterns for Single-Pixel Classification (LIPS-PC). The patterns are learned end-to-end while the measurement operator remains one fixed linear matrix, identical for every input. We parameterize it as a residual correction to a two-dimensional discrete cosine transform (2D-DCT), D=D_base+Δ, and jointly train the correction and a compact classifier using cross-entropy, without a reconstruction objective. This yields a task-driven linear embedding through optimization. We evaluate LIPS-PC numerically on MNIST and Fashion-MNIST under a controlled protocol, then on RIAWELC, a real industrial X-ray weld-radiograph dataset. Across five budgets, M∈8,16,32,64,128, corresponding to sampling ratios from below 1% to 12.5%, the learned operator outperforms the best tested fixed transform by 9-15 percentage points at M=8 on all three datasets. Its eight-measurement accuracy requires roughly 16-32 measurements with the best fixed transform, indicating a 2× to 4× saving at comparable accuracy. The advantage generally shrinks as M increases, with a small reversal on RIAWELC at higher budgets. These results show that learning what to measure is especially valuable when measurements are scarce.
ER -