Optimal Multispectral Imaging using RGB Cameras

Damir Ser\v{s}i\'c; Dubravko Babi\'c; Ivan \v{S}krabo; Tomislav Matuli\'c

arxiv: 2604.19460 · v1 · submitted 2026-04-21 · 📡 eess.SP · eess.IV

Optimal Multispectral Imaging using RGB Cameras

Tomislav Matuli\'c , Ivan \v{S}krabo , Dubravko Babi\'c , Damir Ser\v{s}i\'c This is my paper

Pith reviewed 2026-05-10 01:55 UTC · model grok-4.3

classification 📡 eess.SP eess.IV

keywords multispectral imagingRGB cameraswavelength allocationcondition numberspectral reconstructionlinear measurement modeloptical filtersnoise robustness

0 comments

The pith

Allocating wavelengths across RGB cameras and multi-band filters by minimizing the condition number of the stacked measurement matrix produces the most stable spectral reconstruction.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes that a low-cost multispectral imager can be built by combining ordinary RGB cameras with off-the-shelf narrowband filters, then systematically choosing which target wavelengths each camera-filter pair should capture. It formulates the camera responses as a linear system whose mixing matrix is fixed once the wavelength-to-unit assignments are chosen, and it selects the assignment that yields the smallest condition number of that matrix. A reader should care because the method turns an otherwise open-ended hardware-design task into a finite, exhaustive search that directly controls numerical stability and worst-case noise performance without requiring custom sensors.

Core claim

Wavelength allocation is treated as a deterministic design problem whose solution is the configuration that minimizes the spectral condition number of the global system matrix obtained by stacking the linear measurement models of every camera-filter unit; frame-theoretic considerations show that this choice promotes numerical stability, maximizes the minimum output signal-to-noise ratio, and improves robustness of the recovered spectra, with additional gains available from redundant assignments in which some wavelengths are observed by more than one unit.

What carries the argument

The spectral condition number of the stacked linear measurement matrix whose rows are the products of camera spectral sensitivities and filter transmittances for each assigned wavelength.

If this is right

The finite design space permits exhaustive evaluation of every feasible wavelength-to-camera-filter assignment under given hardware constraints.
Redundant wavelength assignments, in which the same band is captured by multiple camera-filter units, supply extra degrees of freedom that further lower noise sensitivity.
Numerical stability and worst-case signal-to-noise ratio are simultaneously improved by the single scalar criterion of condition-number minimization.
The same framework can be re-run whenever new filter sets or camera models become available, because only the measured sensitivity curves and transmittance curves are required as input.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The approach could be tested on scenes with spatially varying illumination to check whether the stability gains survive real-world lighting changes not captured in the static linear model.
Similar condition-number optimization might be applied to select filter placements in other snapshot spectral systems that also rely on linear demixing.
In practice one would still need to verify that the chosen allocation remains near-optimal after small manufacturing tolerances or temperature-induced shifts in filter curves are included.

Load-bearing premise

The linear measurement model exactly describes real camera responses and the spectral sensitivities together with the filter transmittances are known precisely and remain constant across scenes and angles.

What would settle it

Acquire ground-truth spectra of a calibrated test target under controlled illumination, reconstruct them using both the condition-number-minimizing allocation and a randomly chosen allocation, then compare the root-mean-square reconstruction errors; the optimized allocation must show measurably lower error if the claim holds.

Figures

Figures reproduced from arXiv: 2604.19460 by Damir Ser\v{s}i\'c, Dubravko Babi\'c, Ivan \v{S}krabo, Tomislav Matuli\'c.

**Figure 2.** Figure 2: Spectral performance of the CMOS sensor Sony [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗

**Figure 3.** Figure 3: Optimal spectral distribution of the wavelengths onto [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

read the original abstract

We present a physics-driven framework for accurate evaluation of discrete spectral bands using a low-cost multispectral setup built from off-the-shelf RGB cameras and narrow multi-band optical filters. The approach starts by explicitly formulating a linear measurement model. The camera responses are expressed as linear mixtures of unknown spectral components, with mixing coefficients determined by the overlap between the camera spectral sensitivities and the filter transmittances. For a multi-camera configuration, the per-camera models are stacked into a single global system whose structure is fully determined by the allocation of target wavelengths across the camera--filter units. We pose wavelength allocation as a deterministic design problem and select the configuration that minimizes the spectral condition number of the resulting system matrix. Guided by a frame-theoretic interpretation, this criterion promotes numerical stability, maximizes worst-case output signal-to-noise ratio, and improves the robustness of spectral reconstruction. The design space is finite, enabling the evaluation of all feasible configurations under practical constraints. We demonstrate the method on a representative example with 12 target wavelengths and four triband filters, and identify the wavelength allocation that yields the most stable and noise-robust recovery. The proposed framework includes redundant configurations, in which individual wavelengths are measured by multiple cameras, thereby providing additional degrees of freedom that further improve noise robustness.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper gives a clean combinatorial optimization for assigning wavelengths to RGB camera plus filter units by minimizing the condition number of the stacked measurement matrix, with room for redundant measurements.

read the letter

This paper gives a clean combinatorial optimization for assigning wavelengths to RGB camera plus filter units by minimizing the condition number of the stacked measurement matrix, with room for redundant measurements. The authors start from the standard linear overlap model and treat allocation as a finite search problem whose objective is the spectral condition number. They tie the choice to frame bounds for stability and worst-case SNR, then show one concrete case with 12 target bands and four triband filters. Allowing some bands to be captured by more than one unit adds extra degrees of freedom without new hardware. That combination of exhaustive search plus redundancy is the part that feels fresh within the engineering literature on low-cost multispectral rigs. The model itself is set up directly from the physical integrals, so the math stays transparent and the design space stays enumerable under realistic constraints. Credit for keeping the approach deterministic and parameter-free. The main limitation is the assumption that every camera sensitivity curve and filter transmittance is known exactly and stays fixed. Real devices shift with angle, temperature, and unit variation, so the pre-selected allocation can lose its optimality once the hardware is in the field. The abstract example is described at a high level; without seeing the actual reconstruction errors or a perturbation test it is hard to judge how much robustness is gained over simpler assignments. This work is aimed at applied imaging groups who already build systems from commodity RGB cameras and want a systematic way to pick filter wavelengths. A reader who needs reproducible, low-cost spectral recovery for agriculture or inspection tasks could adapt the method quickly. The thinking is clear and the citations stay on point. It deserves a serious referee because the core idea is executable and the practical payoff is immediate, even if the validation section needs more detail on mismatch robustness.

Referee Report

2 major / 2 minor

Summary. The manuscript presents a physics-driven framework for multispectral imaging with RGB cameras and narrow multi-band filters. It formulates camera responses via a linear model whose mixing coefficients are the overlap integrals between known spectral sensitivities and filter transmittances. For a multi-camera setup these per-unit models are stacked into a global system matrix A whose structure depends on the discrete allocation of target wavelengths to camera-filter pairs. The central design step selects the allocation that minimizes cond(A); this choice is motivated by a frame-theoretic argument that it improves numerical stability, worst-case SNR, and reconstruction robustness. The finite design space permits exhaustive enumeration, which is demonstrated on an example using 12 target wavelengths and four triband filters, including redundant allocations.

Significance. If the central claim holds, the work supplies a deterministic, parameter-free design procedure for low-cost multispectral systems that directly optimizes a well-understood stability metric. The explicit linear model, the exhaustive search over feasible allocations, and the inclusion of redundant measurements are concrete strengths. The frame-theoretic justification for the condition-number criterion adds theoretical grounding that is not always present in heuristic filter-selection papers.

major comments (2)

[§5] §5 (representative example with 12 wavelengths and four triband filters): the claim that the chosen allocation 'yields the most stable and noise-robust recovery' is made with respect to the nominal system matrix A constructed from idealized sensitivity curves. No perturbation analysis or Monte-Carlo evaluation over realistic variations in those curves (incidence angle, temperature, unit-to-unit differences) is reported, so it is unclear whether the reported cond(A) minimum remains near-optimal under the mismatches that occur in physical hardware.
[Linear measurement model] Linear measurement model (prior to the optimization section): the global matrix A is assembled by stacking products of fixed camera sensitivities and filter transmittances. The optimality argument therefore inherits any modeling error in those curves; because the manuscript treats the curves as known exactly, the predicted robustness gains are guaranteed only under that assumption and may not materialize when the actual responses deviate.

minor comments (2)

The abstract invokes a 'frame-theoretic interpretation' without a short derivation or citation to the specific frame bounds that link cond(A) to worst-case SNR; adding one or two sentences would improve accessibility.
Notation for the stacked system matrix A and the per-camera sub-matrices should be introduced with explicit equation numbers so that later references to 'the resulting system matrix' are unambiguous.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on the robustness of the proposed wavelength-allocation design under realistic deviations from the idealized spectral curves. We address each major comment below and will incorporate additional analysis and discussion in the revised manuscript.

read point-by-point responses

Referee: [§5] §5 (representative example with 12 wavelengths and four triband filters): the claim that the chosen allocation 'yields the most stable and noise-robust recovery' is made with respect to the nominal system matrix A constructed from idealized sensitivity curves. No perturbation analysis or Monte-Carlo evaluation over realistic variations in those curves (incidence angle, temperature, unit-to-unit differences) is reported, so it is unclear whether the reported cond(A) minimum remains near-optimal under the mismatches that occur in physical hardware.

Authors: We agree that the optimality claim in §5 is established only for the nominal matrix A derived from idealized curves, and that a perturbation study is needed to assess whether the selected allocation remains preferable under realistic mismatches. In the revision we will add a Monte-Carlo experiment that applies bounded perturbations (e.g., ±5 % amplitude scaling, ±2 nm center-wavelength shifts, and small incidence-angle-induced changes) to the sensitivity and transmittance curves, recomputes cond(A) for the top-ranked allocations, and reports the fraction of trials in which the originally chosen configuration stays within the lowest 10 % of condition numbers. This will quantify the practical stability of the design choice. revision: yes
Referee: [Linear measurement model] Linear measurement model (prior to the optimization section): the global matrix A is assembled by stacking products of fixed camera sensitivities and filter transmittances. The optimality argument therefore inherits any modeling error in those curves; because the manuscript treats the curves as known exactly, the predicted robustness gains are guaranteed only under that assumption and may not materialize when the actual responses deviate.

Authors: The linear model is formulated under the assumption that the camera sensitivities and filter transmittances are known to sufficient accuracy (via manufacturer data or per-unit calibration). The condition-number criterion then guarantees improved numerical stability and worst-case SNR within that model. We will revise the manuscript to state this modeling assumption explicitly in the linear-measurement-model section and to add a short paragraph noting that, in hardware, calibration measurements should be used to construct A. The perturbation analysis described above will also serve to illustrate the sensitivity of the optimality result to curve mismatch. revision: yes

Circularity Check

0 steps flagged

No significant circularity in wavelength allocation optimization

full rationale

The paper constructs the system matrix A explicitly from the known overlap integrals between camera spectral sensitivities and filter transmittances at chosen target wavelengths, then selects the allocation minimizing cond(A) via exhaustive enumeration of the finite discrete design space. This is a standard, non-circular optimization of a well-defined objective function derived directly from the linear measurement model; the result does not reduce to any input by construction, nor does it rely on fitted parameters, self-referential predictions, or load-bearing self-citations. The frame-theoretic interpretation is presented only as interpretive guidance for why the condition-number criterion is desirable, not as a mathematical step that forces the outcome.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The framework rests on the standard linear imaging model and known device characteristics; no new free parameters, ad-hoc axioms, or invented entities are introduced in the abstract.

axioms (1)

domain assumption Camera responses are linear mixtures of unknown spectral components with mixing coefficients given by the overlap of camera sensitivities and filter transmittances
Explicitly stated as the starting point of the measurement model.

pith-pipeline@v0.9.0 · 5537 in / 1264 out tokens · 45979 ms · 2026-05-10T01:55:20.737465+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

18 extracted references · 18 canonical work pages

[1]

A review of the application of uav multispectral remote sensing technology in precision agriculture,

S. Zhang, X. Wang, H. Lin, Y . Dong, and Z. Qiang, “A review of the application of uav multispectral remote sensing technology in precision agriculture,”Smart Agricultural Technology, vol. 12, p. 101406, Dec

work page
[2]

Available: http://dx.doi.org/10.1016/j.atech.2025.101406

[Online]. Available: http://dx.doi.org/10.1016/j.atech.2025.101406

work page doi:10.1016/j.atech.2025.101406 2025
[3]

Multispectral imaging for skin diseases assessment—state of the art and perspectives,

M.-A. Ilis ,anu, F. Moldoveanu, and A. Moldoveanu, “Multispectral imaging for skin diseases assessment—state of the art and perspectives,” Sensors, vol. 23, no. 8, p. 3888, Apr. 2023. [Online]. Available: http://dx.doi.org/10.3390/s23083888

work page doi:10.3390/s23083888 2023
[4]

Multispectral imaging for plant food quality analysis and visualization,

W. Su and D. Sun, “Multispectral imaging for plant food quality analysis and visualization,”Comprehensive Reviews in Food Science and Food Safety, vol. 17, no. 1, p. 220–239, Jan. 2018. [Online]. Available: http://dx.doi.org/10.1111/1541-4337.12317

work page doi:10.1111/1541-4337.12317 2018
[5]

Hyperspectral and multispectral imaging for evaluating food safety and quality,

J. Qin, K. Chao, M. S. Kim, R. Lu, and T. F. Burks, “Hyperspectral and multispectral imaging for evaluating food safety and quality,”Journal of Food Engineering, vol. 118, no. 2, p. 157–171, Sep. 2013. [Online]. Available: http://dx.doi.org/10.1016/j.jfoodeng.2013.04.001

work page doi:10.1016/j.jfoodeng.2013.04.001 2013
[6]

Understanding multispectral imaging of cultural heritage: Determining best practice in msi analysis of historical artefacts,

C. Jones, C. Duffy, A. Gibson, and M. Terras, “Understanding multispectral imaging of cultural heritage: Determining best practice in msi analysis of historical artefacts,”Journal of Cultural Heritage, vol. 45, p. 339–350, Sep. 2020. [Online]. Available: http://dx.doi.org/1 0.1016/j.culher.2020.03.004

work page 2020
[7]

Six band hdtv camera system for spectrum-based color reproduction,

K. Ohsawa, T. Ajito, Y . Komiya, H. Fukuda, H. Haneishi, M. Yamaguchi, and N. Ohyama, “Six band hdtv camera system for spectrum-based color reproduction,”Journal of Imaging Science and Technology, vol. 48, no. 2, p. 85–92, Mar. 2004. [Online]. Available: http://dx.doi.org/10.2352/J.ImagingSci.Technol.2004.48.2.art00003

work page doi:10.2352/j.imagingsci.technol.2004.48.2.art00003 2004
[8]

PoseNet: A convolutional network for real-time 6-dof camera relocalization,

J.-I. Park, M.-H. Lee, M. D. Grossberg, and S. K. Nayar, “Multispectral imaging using multiplexed illumination,” in2007 IEEE 11th International Conference on Computer Vision. IEEE, 2007, p. 1–8. [Online]. Available: http://dx.doi.org/10.1109/ICCV .2007.4409090

work page doi:10.1109/iccv 2007
[9]

Multispectral imaging using a stereo camera: concept, design and assessment,

R. Shrestha, A. Mansouri, and J. Y . Hardeberg, “Multispectral imaging using a stereo camera: concept, design and assessment,”EURASIP Journal on Advances in Signal Processing, vol. 2011, no. 1, Sep. 2011. [Online]. Available: http://dx.doi.org/10.1186/1687-6180-2011-57

work page doi:10.1186/1687-6180-2011-57 2011
[10]

A comparative analysis of spectral reflectance estimated in various spaces using a trichromatic camera system,

F. H. Imai, R. S. Berns, and D.-Y . Tzeng, “A comparative analysis of spectral reflectance estimated in various spaces using a trichromatic camera system,”Journal of Imaging Science and Technology, vol. 44, no. 4, p. 280–287, Jul. 2000. [Online]. Available: http://dx.doi.org/10.2352/J.ImagingSci.Technol.2000.44.4.art00003

work page doi:10.2352/j.imagingsci.technol.2000.44.4.art00003 2000
[11]

Unsupervised spectral reconstruction from rgb images under two lighting conditions,

X. Cao, Y . Lian, Z. Liu, J. Li, and K. Wang, “Unsupervised spectral reconstruction from rgb images under two lighting conditions,”Optics Letters, vol. 49, no. 8, p. 1993, Apr. 2024. [Online]. Available: http://dx.doi.org/10.1364/OL.517007

work page doi:10.1364/ol.517007 1993
[12]

Spectral imaging with deep learning,

L. Huang, R. Luo, X. Liu, and X. Hao, “Spectral imaging with deep learning,”Light: Science & Applications, vol. 11, no. 1, Mar. 2022. [Online]. Available: http://dx.doi.org/10.1038/s41377-022-00743-6

work page doi:10.1038/s41377-022-00743-6 2022
[13]

Filter selection for multispectral color image acquisition,

J. Y . Hardeberg, “Filter selection for multispectral color image acquisition,” inImage Processing, Image Quality, Image Capture Systems Conference, 2004. [Online]. Available: https://api.semanticscho lar.org/CorpusID:32623590

work page 2004
[14]

Filter selection for optimizing the spectral sensitivity of broadband multispectral cameras based on maximum linear independence,

S.-X. Li, “Filter selection for optimizing the spectral sensitivity of broadband multispectral cameras based on maximum linear independence,”Sensors, vol. 18, no. 5, p. 1455, May 2018. [Online]. Available: http://dx.doi.org/10.3390/s18051455

work page doi:10.3390/s18051455 2018
[15]

Optimiz- ing Federated Unsupervised Person Re-identification via Camera-aware Clustering,

F. Sippel, J. Seiler, and A. Kaup, “Optimal filter selection for multispectral object classification using fast binary search,” in 2022 IEEE 24th International Workshop on Multimedia Signal Processing (MMSP). IEEE, Sep. 2022, p. 1–5. [Online]. Available: http://dx.doi.org/10.1109/MMSP55362.2022.9949059

work page doi:10.1109/mmsp55362.2022.9949059 2022
[16]

Camera array for multi-spectral imaging,

N. Genser, J. Seiler, and A. Kaup, “Camera array for multi-spectral imaging,”IEEE Transactions on Image Processing, vol. 29, p. 9234–9249, 2020. [Online]. Available: http://dx.doi.org/10.1109/TIP.2 020.3024738

work page doi:10.1109/tip.2 2020
[17]

Multispectral imaging using multiple-bandpass filters,

G. Themelis, J. S. Yoo, and V . Ntziachristos, “Multispectral imaging using multiple-bandpass filters,”Opt. Lett., vol. 33, no. 9, pp. 1023–1025, May 2008. [Online]. Available: https://opg.optica.org/ol/ab stract.cfm?URI=ol-33-9-1023

work page 2008
[18]

Implementation - optimal multispectral imaging using rgb cameras,

I. ˇSkrabo, T. Matuli ´c, D. Babi ´c, and D. Ser ˇsi´c, “Implementation - optimal multispectral imaging using rgb cameras,” 2026. [Online]. Available: https://www.zesoi.fer.hr/ download/repository/OptimalMultis pectralImagingUsingRGBCameras.zip

work page 2026

[1] [1]

A review of the application of uav multispectral remote sensing technology in precision agriculture,

S. Zhang, X. Wang, H. Lin, Y . Dong, and Z. Qiang, “A review of the application of uav multispectral remote sensing technology in precision agriculture,”Smart Agricultural Technology, vol. 12, p. 101406, Dec

work page

[2] [2]

Available: http://dx.doi.org/10.1016/j.atech.2025.101406

[Online]. Available: http://dx.doi.org/10.1016/j.atech.2025.101406

work page doi:10.1016/j.atech.2025.101406 2025

[3] [3]

Multispectral imaging for skin diseases assessment—state of the art and perspectives,

M.-A. Ilis ,anu, F. Moldoveanu, and A. Moldoveanu, “Multispectral imaging for skin diseases assessment—state of the art and perspectives,” Sensors, vol. 23, no. 8, p. 3888, Apr. 2023. [Online]. Available: http://dx.doi.org/10.3390/s23083888

work page doi:10.3390/s23083888 2023

[4] [4]

Multispectral imaging for plant food quality analysis and visualization,

W. Su and D. Sun, “Multispectral imaging for plant food quality analysis and visualization,”Comprehensive Reviews in Food Science and Food Safety, vol. 17, no. 1, p. 220–239, Jan. 2018. [Online]. Available: http://dx.doi.org/10.1111/1541-4337.12317

work page doi:10.1111/1541-4337.12317 2018

[5] [5]

Hyperspectral and multispectral imaging for evaluating food safety and quality,

J. Qin, K. Chao, M. S. Kim, R. Lu, and T. F. Burks, “Hyperspectral and multispectral imaging for evaluating food safety and quality,”Journal of Food Engineering, vol. 118, no. 2, p. 157–171, Sep. 2013. [Online]. Available: http://dx.doi.org/10.1016/j.jfoodeng.2013.04.001

work page doi:10.1016/j.jfoodeng.2013.04.001 2013

[6] [6]

Understanding multispectral imaging of cultural heritage: Determining best practice in msi analysis of historical artefacts,

C. Jones, C. Duffy, A. Gibson, and M. Terras, “Understanding multispectral imaging of cultural heritage: Determining best practice in msi analysis of historical artefacts,”Journal of Cultural Heritage, vol. 45, p. 339–350, Sep. 2020. [Online]. Available: http://dx.doi.org/1 0.1016/j.culher.2020.03.004

work page 2020

[7] [7]

Six band hdtv camera system for spectrum-based color reproduction,

K. Ohsawa, T. Ajito, Y . Komiya, H. Fukuda, H. Haneishi, M. Yamaguchi, and N. Ohyama, “Six band hdtv camera system for spectrum-based color reproduction,”Journal of Imaging Science and Technology, vol. 48, no. 2, p. 85–92, Mar. 2004. [Online]. Available: http://dx.doi.org/10.2352/J.ImagingSci.Technol.2004.48.2.art00003

work page doi:10.2352/j.imagingsci.technol.2004.48.2.art00003 2004

[8] [8]

PoseNet: A convolutional network for real-time 6-dof camera relocalization,

J.-I. Park, M.-H. Lee, M. D. Grossberg, and S. K. Nayar, “Multispectral imaging using multiplexed illumination,” in2007 IEEE 11th International Conference on Computer Vision. IEEE, 2007, p. 1–8. [Online]. Available: http://dx.doi.org/10.1109/ICCV .2007.4409090

work page doi:10.1109/iccv 2007

[9] [9]

Multispectral imaging using a stereo camera: concept, design and assessment,

R. Shrestha, A. Mansouri, and J. Y . Hardeberg, “Multispectral imaging using a stereo camera: concept, design and assessment,”EURASIP Journal on Advances in Signal Processing, vol. 2011, no. 1, Sep. 2011. [Online]. Available: http://dx.doi.org/10.1186/1687-6180-2011-57

work page doi:10.1186/1687-6180-2011-57 2011

[10] [10]

A comparative analysis of spectral reflectance estimated in various spaces using a trichromatic camera system,

F. H. Imai, R. S. Berns, and D.-Y . Tzeng, “A comparative analysis of spectral reflectance estimated in various spaces using a trichromatic camera system,”Journal of Imaging Science and Technology, vol. 44, no. 4, p. 280–287, Jul. 2000. [Online]. Available: http://dx.doi.org/10.2352/J.ImagingSci.Technol.2000.44.4.art00003

work page doi:10.2352/j.imagingsci.technol.2000.44.4.art00003 2000

[11] [11]

Unsupervised spectral reconstruction from rgb images under two lighting conditions,

X. Cao, Y . Lian, Z. Liu, J. Li, and K. Wang, “Unsupervised spectral reconstruction from rgb images under two lighting conditions,”Optics Letters, vol. 49, no. 8, p. 1993, Apr. 2024. [Online]. Available: http://dx.doi.org/10.1364/OL.517007

work page doi:10.1364/ol.517007 1993

[12] [12]

Spectral imaging with deep learning,

L. Huang, R. Luo, X. Liu, and X. Hao, “Spectral imaging with deep learning,”Light: Science & Applications, vol. 11, no. 1, Mar. 2022. [Online]. Available: http://dx.doi.org/10.1038/s41377-022-00743-6

work page doi:10.1038/s41377-022-00743-6 2022

[13] [13]

Filter selection for multispectral color image acquisition,

J. Y . Hardeberg, “Filter selection for multispectral color image acquisition,” inImage Processing, Image Quality, Image Capture Systems Conference, 2004. [Online]. Available: https://api.semanticscho lar.org/CorpusID:32623590

work page 2004

[14] [14]

Filter selection for optimizing the spectral sensitivity of broadband multispectral cameras based on maximum linear independence,

S.-X. Li, “Filter selection for optimizing the spectral sensitivity of broadband multispectral cameras based on maximum linear independence,”Sensors, vol. 18, no. 5, p. 1455, May 2018. [Online]. Available: http://dx.doi.org/10.3390/s18051455

work page doi:10.3390/s18051455 2018

[15] [15]

Optimiz- ing Federated Unsupervised Person Re-identification via Camera-aware Clustering,

F. Sippel, J. Seiler, and A. Kaup, “Optimal filter selection for multispectral object classification using fast binary search,” in 2022 IEEE 24th International Workshop on Multimedia Signal Processing (MMSP). IEEE, Sep. 2022, p. 1–5. [Online]. Available: http://dx.doi.org/10.1109/MMSP55362.2022.9949059

work page doi:10.1109/mmsp55362.2022.9949059 2022

[16] [16]

Camera array for multi-spectral imaging,

N. Genser, J. Seiler, and A. Kaup, “Camera array for multi-spectral imaging,”IEEE Transactions on Image Processing, vol. 29, p. 9234–9249, 2020. [Online]. Available: http://dx.doi.org/10.1109/TIP.2 020.3024738

work page doi:10.1109/tip.2 2020

[17] [17]

Multispectral imaging using multiple-bandpass filters,

G. Themelis, J. S. Yoo, and V . Ntziachristos, “Multispectral imaging using multiple-bandpass filters,”Opt. Lett., vol. 33, no. 9, pp. 1023–1025, May 2008. [Online]. Available: https://opg.optica.org/ol/ab stract.cfm?URI=ol-33-9-1023

work page 2008

[18] [18]

Implementation - optimal multispectral imaging using rgb cameras,

I. ˇSkrabo, T. Matuli ´c, D. Babi ´c, and D. Ser ˇsi´c, “Implementation - optimal multispectral imaging using rgb cameras,” 2026. [Online]. Available: https://www.zesoi.fer.hr/ download/repository/OptimalMultis pectralImagingUsingRGBCameras.zip

work page 2026