Virtually structured illumination for terahertz

super-resolution imaging
J AMES P. F LEMING 1, 2, * , L UCY A. D OWNES 2 , J OHN M. G IRKIN 1 AND
K EVIN J. W EATHERILL 2
1 Centrefor Advanced Instrumentation, Department of Physics, Durham University, South Road, Durham
DH1 3LE, United Kingdom
arXiv:2504.12092v1 [physics.optics] 16 Apr 2025

2 Quantum Light and Matter Group, Department of Physics, Durham University, South Road, Durham DH1
3LE, United Kingdom
* [email protected]

Abstract: We demonstrate structured illumination super-resolution imaging in the Terahertz

(THz) frequency band using the Virtually Structured Detection (VSD) method. Leveraging
our previously reported high-speed, high-sensitivity atomic-based THz imager, we achieve a
resolution enhancement of (74 ± 3)% at 0.55 THz, without the aid of deconvolution methods.
We show a high-speed THz imaging system is compatible with the use of advanced optical
techniques, with potential disruptive effects on applications requiring both high speed and high
spatial resolution imaging in the THz range.

1. Introduction
Imaging in the THz band (0.3-3 THz) [1] has many applications ranging from security screening [2]
and non-destructive testing [3] to biomedical diagnostics [4]. These applications take advantage
of the favorable properties of the THz frequency range, such as its ability to penetrate common
non-conducting materials like wood, paper, and plastics [5]. Additionally, its non-ionizing nature
makes it a safer alternative to higher-frequency penetrating radiation, further broadening its
appeal across various domains.
THz imaging can broadly be categorized into two general schemes: single-point (SP) detection
and focal-plane array (FPA) detection. While SP detectors typically acquire images by raster-
scanning a target object, FPA detectors can capture images in a single shot using a 2D array
of sensors for parallel acquisition. Through the use of far-field aperture-based raster scanning
techniques, SP detectors can achieve sub-diffraction limited spatial resolution and, when combined
with time-domain spectroscopy (TDS), can produce images with hyperspectral information.
However, the raster-scanning process is inherently slow, making SP detection unsuitable for
applications requiring rapid image acquisition. In contrast, FPA detectors offer significantly faster
imaging speeds by capturing images in a single-shot. Recent advancements in room-temperature
FPAs have led to commercially available THz cameras with video-rate acquisition performance.
However, despite these advances, THz FPA detectors generally suffer from low sensitivity, high
noise, provide no spectral information, and ultimately are constrained by diffraction-limited
spatial resolution.
Techniques to overcome the diffraction limit have been investigated in optical microscopy
for many years. One example of such super-resolution techniques is Structured Illumination
Microscopy (SIM) [6, 7]. SIM involves illuminating a sample with a known spatial pattern to
allow the extraction of high-frequency spatial information. This technique can increase spatial
resolution of images by up to a factor of two above the diffraction limit [8]. One downside of
SIM is that it requires the acquisition of numerous images in order to collect sufficient spatial
information. Whilst this is not a major issue for optical imaging where sensors are fast and
have low noise, it is a more significant problem in the THz range and previous attempts of
super-resolution imaging at THz frequencies have either been purely through computational
enhancement [9, 10] or demonstrated via SP detection [11].
 Recently, a novel THz FPA sensor based on THz-to-optical conversion in room temperature,
optically pumped atomic vapor has emerged [12, 13]. This technique has demonstrated very high
image acquisition rates in comparison to other room-temperature FPA sensors, and because images
are captured using standard optical cameras, sensors can be chosen to match the application.
Here, we make use of a fast atom-based THz imaging system to apply a variant of SIM
known as Virtually Structured Detection (VSD) [14, 15] to achieve THz super-resolution imaging.
Being able to perform super-resolution imaging at 0.55 THz holds promise for applications that
require high-speed and high-spatial resolution, while leveraging the deep penetrating properties
of 0.55 THz radiation.

2. Virtually Structured Detection

SIM often relies on complex illumination systems that are impractical at THz frequencies [16–18].
VSD simplifies this by using a point or line illumination to scan the object, capturing a full-field
image at each scan position. For THz VSD, we employ a scanning line illumination [14, 15, 19],
enabling low-complexity THz super-resolution imaging. VSD reconstruction begins by taking
each image scan, multiplying it by a modulation mask 𝑚(r), and spatially integrating it along
the scan axis - reducing it to a single-pixel-wide image [14]. This processed scan 𝑝 𝑖 (ri ) is
represented as

𝑝 𝑖 (ri ) = ℎ𝑖𝑙 (r) ⊗ {[ℎ 𝑑𝑒 (r) ⊗ 𝑚(r)] 𝑠(r)} , (1)

where r ≡ (𝑥, 𝑦) is the spatial position vector, 𝑠(r) the sample object, ℎ𝑖𝑙 (r) and ℎ 𝑑𝑒 (r) are the
point spread functions (PSF) of the respective illumination and detection paths of the imaging
system, and ⊗ represents the image convolution operator. Successive scans are then stacked
to generate a virtually structured image 𝑝(r) = [ 𝑝 1 (r1 ) · · · 𝑝 𝑛 (rn )], which is mathematically
equivalent to those acquired in conventional widefield SIM [19]. As with widefield SIM, a
sinusoidal modulation 𝑚(r) = cos (2𝜋p 𝜃 · r + 𝜙) is employed, where p 𝜃 ≡ (p cos 𝜃, p sin 𝜃)
with 𝜃 being the scan angle and 𝜙 the modulations phase. The significance of such modulation
becomes clearer from its Fourier transform, 𝑀˜ (k):

𝑀˜ (k) = 𝛿 (k + p 𝜃 ) 𝑒 𝑖 𝜙 + 𝛿 (k − p 𝜃 ) 𝑒 −𝑖 𝜙 . (2)
Here, the modulation comprises of two delta functions, 𝛿 (k ± p 𝜃 ), which when convolved
˜
with the Fourier spectrum of the object, 𝑆(k), shifts the content of the spectrum by the spatial
frequency vector, ±p 𝜃 . This is a feature of the Fourier shifting property [20]. Under such
˜
modulation 𝑃(k), the Fourier transform of 𝑝(r), is realized as

˜ = 𝐻˜ (k) 𝑆(k ˜ − p 𝜃 )𝑒 −𝑖 𝜙 .
˜ + p 𝜃 )𝑒 𝑖 𝜙 + 𝑆(k
 
𝑃(k) (3)
These shifted Fourier bands 𝑆(k ˜ ± p 𝜃 ) contain higher spatial frequency information that
otherwise would be cut-off by the optical transfer function (OTF) of the imaging system
˜ ± p 𝜃 ) by changing the
𝐻˜ (k) = 𝐻˜ 𝑖𝑙 (k) 𝐻˜ 𝑑𝑒 (k). Solving Eq. (3) for the shifted Fourier bands 𝑆(k
phase of the modulation, 𝜙, allows for retrieval of the two shifted bands,

˜ ± p 𝜃 ) = 𝑃(k,
𝐻˜ (k) 𝑆(k ˜ 𝜙 = 0) ± 𝑖 𝑃(k,
˜ 𝜙 = 𝜋/2) . (4)
As the modulation is applied computationally, no additional image scans are required for the
additional phase. Unlike SIM, the reconstruction is phase artifact-free as the phase is exactly
defined [19]. Typically, attenuation of spatial frequencies caused by 𝐻˜ (k) can be restored via a
final deconvolution step

˜ ± p 𝜃 ) = deconvolve 𝑃(k,
˜ 𝜙 = 0) ± 𝑖 𝑃(k,
˜ 𝜙 = 𝜋/2) .

𝑆(k (5)
 For this work, we opted not to deconvolve our images to assess the standalone performance of
the VSD technique. We direct the reader to the extensive use of various deconvolution algorithms
in the literature [21–24].
Super-resolution image reconstruction concludes by summation of the two shifted Fourier
bands, improving the spatial frequency extent for a theoretical maximum two-fold resolution
when |p 𝜃 | = kcutoff , the cut-off spatial frequency of the widefield system [25]. Such improvement
is observed across the scan-axis only. For near-isotropic resolution improvement, the angle over
which the sample is scanned is varied, typically as 𝜃 = 0°, 60° and 120°.

3. Experimental Implementation

Fig. 1. a) Energy ladder scheme in Cesium used for the conversion of 0.55 THz to
the green 535 nm via decay fluorescence. b) True-color composite image of the vapor
cell with 1 cm2 sensing area, both unfiltered background fluorescence and green signal
fluorescence are shown. c) Processed monochromatic THz image from the acquired
fluorescence signal.

Widefield THz images were collected using our atom-based imaging system, full details
of which are available in previous publications [13, 26]. To summarize, using infrared lasers,
frequency stabilized to atomic references [27, 28], we can promote alkali metals to highly excited
states, known as Rydberg states [29]. Rydberg states feature electric-dipole transitions that fall
within the THz frequency band and are therefore sensitive to resonant THz fields. Rydberg
states also spontaneously decay via probabilistic decay pathways back to the atomic ground state,
producing an optical fluorescence spectra. By illuminating a thermal vapor of optically pumped
alkali metal atoms with a resonant THz-field, the appropriate Rydberg transition can be driven,
which will then decay to produce a detectable optical fluorescence signal. By shaping each laser
with a combination of cylindrical lenses, overlapping lightsheets can be formed, constructing an
active sensing focal-plane. High-speed, high-sensitivity focal-plane THz imaging is therefore
realized by efficient THz-to-optical conversion. The sub-microsecond decay rates of the Rydberg
states allows theoretical imaging rates in the 1 MHz range. In practice the imaging frame-rate
is limited by the signal-to-noise and available throughput of the optical readout system, which
has been demonstrated to a maximum of 12,000 frames per second, an almost three-orders of
magnitude improvement when compared to the state-of-the-art commercial THz cameras [26].
In this study, we operate our Rydberg-based THz imaging system at 0.55 THz by using the
14P3/2 → 13D5/2 transition in Cesium, as shown in Fig. 1a. A Cesium thermal vapor (Fig. 1b)
is contained in a heated quartz cell and maintained at 45◦ , which maximizes the fluorescence
 THz Sensi�ve
Op�cal
Imaging Lens Atom Imager
Readout
Object mask

Lens relay
Line mask

f Linescan
Images
f
Phase
f Images
f
Collimated THz
Beam Input
vs. Digital modula�on
+ stacking
Fourier
Widefield Super-resolved reconstruc�on
Image Image

Fig. 2. Diagrammatic overview of the THz VSD implementation, showing the

working elements required to generate full-frame image scans. Our atomic imager
facilitates efficient THz-to-optical conversion for fast, full-frame optical readout via a
monochromatic CCD (not shown). To generate a virtually structured image, successive
images are taken as the object is translated through the THz line illumination. These
are digitally modulated, spatially integrated and stacked to produce the phase images,
which are used to reconstruct the super-resolve image.

signal without the thermal vapor becoming optically thick to the infrared excitation lasers. The
vapor cell provides optical access on all four sides. Three frequency-stablized infrared lasers
(𝜆 = 852 nm, 1470 nm, 843 nm) drive the three-step excitation scheme and prepare Cesium in
the 14P3/2 state. The size of the vapor cell and available laser power produces an active sensing
region of 1 cm2 . Narrow-band, resonant 550 GHz continuous-wave THz illumination is provided
by a Virginia Diodes Inc. amplifier multiplier chain (AMC). A stable microwave generator seeds
the AMC, and the frequency is upconverted by a factor of 36, providing a maximum power output
of 5 mW. A power output of 0.3 mW was used to illuminate the target without saturating the
imaging system.
To implement VSD, the required line-profile illumination was generated via a mask with a
sub-diffraction, 250 𝜇m wide slit. The illuminated mask was then imaged onto the object plane
using an image relay of two aspheric PTFE plano-convex lenses (d = 50.8 mm, f = 75 mm),
as shown in Fig. 2. The mask was mounted on a manual rotation stage in order to set the scan
angle 𝜃. Unlike previous optical implementations [14, 15, 19], the scanning process occurs by
translating the object normal to the fixed line illumination, using a pair of translation stages
(Thorlabs MTS50-Z8). A transmissive mask was used as the object of interest to image, placed
at the output plane of the image relay. Both illumination and object masks were produced by
milling copper-clad FR4 board. A common material used in electronic circuit board production,
the copper foil cladding blocks transmission, while FR4 substrate is transmissive at 0.55 THz.
Widefield full-frame images of the object are taken using our atomic THz imager. A commercial
1x silicon objective lens (i2S, d = 60 mm, f = 70 mm) images onto the THz-sensitive atomic
vapor. The optical fluorescence from the vapor was captured by an Andor iXon Ultra 888
EMCCD camera, under a 200 ms exposure. An appropriate bandpass filter was used to isolate
 the signal from superfluous background fluorescence. Images are post-processed by subtracting
a background frame to further isolate the signal [26]. Each captured image was then cropped
to a region of interest (ROI) around the line-center of the illumination. A ROI with width of
the expected Airy disc size was found to be an adequate trade-off between signal-to-noise and
rejection of unwanted diffracted illumination. To generate comparative true widefield images, the
illumination mask was removed and the object mask centered to produce an equivalent uniformly
illuminated, single-shot widefield image.

4. Results

+RUL]RQWDO3RVLWLRQPP
  
$ % ( 

+RUL]RQWDO&URVV6HFWLRQV

1RUPDOL]HG,QWHQVLW\

:LGHILHOG




PP 

& ' ) 

9HUWLFDO&URVV6HFWLRQV

1RUPDOL]HG,QWHQVLW\

6XSHU5HVROXWLRQ




PP 
,PDJH6SDFH )RXULHU6SDFH   
9HUWLFDO3RVLWLRQPP

Fig. 3. Widefield (A) and super-resolution (C) images of a USAF resolution target with
their respective Fourier spectra (B and D). Cross-sections of the vertical bars (E) of
Element 2 highlight resolution improvement along the horizontal scan direction only.
The indicated purple region marks the edge used for quantitative analysis resolution
improvement.

To validate the VSD method, Elements 1 and 2 of Group 0 on a USAF resolution target were
imaged in both the widefield and VSD modality. The imaged elements correspond to a lines-per-
millimeter spacing of 1.00 lp/mm and 1.12 lp/mm respectively. For VSD reconstruction, the
target was scanned along the horizontal direction, to produce an improved resolution along the
horizontal only. A total of 48 scan images were taken, leading to a reconstructed super-resolution
image size of 48 × 48 pixels. This size was chosen such that the super-resolution image meets
the Nyquist sampling limit without oversampling [30]. Fig. 3 presents a comparison of widefield
(A) and super-resolution (C) imaging results, including cross-sectional profiles of the vertical (E)
and horizontal (F) bars of Element 2. The cross-section of the vertical bars (E) shows successful
super-resolution, with the vertical bars being well resolved in the super-resolution image (C).
In contrast, the widefield image (A) fails to resolve the vertical bars, as their spatial frequency
exceeds the calculated cut-off of the imaging lens. In contrast, the cross-section of the horizontal
bars (F) for the super-resolution and widefield image show the horizontal bars are unresolved in
both. This is consistent with the resolution enhancement being limited to the horizontal axis.
The widefield Fourier spectrum (B) and super-resolution Fourier spectrum (D) is also shown,
with contours showing the extent of each Fourier band. The extent of the central Fourier bands
in both images is determined by the spatial cut-off frequency of the imaging system. For the
super-resolution Fourier spectrum, two additional recovered Fourier bands (Equation 4) improve
the extent of Fourier spectra along the horizontal direction. This improved extent results in the
the improvement in image resolution.
To quantify the improvement in resolution, the edge profile of one of the horizontal bars in
Element 2 was analyzed. This is highlighted by the purple shaded region in (E). This edge is
representative of the Edge Spread Function (ESF) of the imaging system, which is the cumulative
distribution of the system’s Point Spread Function (PSF) [31]. We characterize the ESF of both
the widefield and super-resolution image by fitting the Gaussian error function erf, and taking the
derivative to produce a Gaussian approximation of the PSF. This resulted in a PSF full width half
maximum of 0.94 ± 0.01 mm for the widefield image and 0.54 ± 0.01 mm for the super-resolution
image. This represents a (74 ± 3)% improvement in resolution. This resolution enhancement is
lower than the theoretically possible factor of two and is limited by the presence of image noise
and attenuation of high spatial frequencies in the Fourier domain. Both factors, however, can be
corrected for through the omitted deconvolution step [25].
To show VSD can be used to produce isotropic resolution improvement, two pictographic
targets were imaged; the Durham University shield and the Greek letter Ψ. Both image icons
measured 8.5 × 9.7 mm to maximize the coverage across the available field-of-view of the
THz imager. The respective widefield and super-resolution images are shown in Fig. 4 A, B,
E and F. The qualitative image masks were scanned along three equally spaced axes at angle
𝜃 = 0°, 60°, 120°, this was achieved by rotating the illumination mask and translating the image
mask normal to the illumination. Each scan axis comprised of a total of 48 scan images and
provides resolution improvement along its axis. Therefore 144 full-frame images were required
to produce one super-resolution image of a 10 × 10 mm field-of-view. In the Fourier domain, the
super-resolution images are constructed from seven distinct Fourier bands; two shifted bands
per scan axis, plus one central band (Fig. 4 D, H). To improve signal-to-noise, the central band
is the mean average of the recovered central bands from the three scans. The seven distinct
Fourier bands are combined to produce near-isotropic extension of the Fourier spectra. For all
Fourier domain images (Fig. 4 C, D, G, H), each Fourier band is highlighted by contours, and
as with Fig. 3 the extent of contour is determined by the calculated spatial cut-off frequency of
the imaging system. By comparison, the super-resolution images exhibit better resolved detail
and high image contrast along edges, this is due to the improved extent of the respective Fourier
spectra from the VSD reconstruction process.

5. Conclusion
This work has demonstrated experimental implementation of structured illumination super-
resolution imaging in the THz band using virtually structured detection, producing a quantitative
resolution improvement of (74 ± 3)%. While less than the theoretical maximum of two-fold
resolution improvement, this has been demonstrated without the typical deconvolution step that
aides recovery of higher spatial frequencies. While SIM-inspired imaging has been considered in
the neighboring microwave band [32], to our knowledge, this is the longest wavelength at which
such technique has been demonstrated experimentally.
A low complexity implementation was prioritized over imaging acquisition speed, in order
to evaluate the VSD technique. While the VSD method is inherently slower than widefield
SIM due to the requirement of more acquired images per reconstruction, the high-speed nature
of our atomic THz imager meant overall acquisition speed was limited by the translation rate
of the object. A faster acquisition scheme could be achieved using a beam-scanning system,
 A 1 mm B 1 mm E 1 mm F 1 mm

C D G H

(a) (b)

Fig. 4. Comparison of true wide-field images (A, E) and super-resolution images (B, F)
of two transmission targets a) the Durham University shield b) the Greek letter Ψ. Each
image is accompanied by its respective Fourier spectrum, demonstrating an increased
spatial frequency content in the super-resolution images.

vastly increasing the acquisition rates at the expense of increased operational complexity. This
would offer video-rate THz imaging at high spatial resolution while utilizing the advantageous
penetrating properties of 0.55 THz through materials, such as FR4 circuit board, that are opaque
at higher THz frequencies.

6. Acknowledgements
We thank Andrew MacKellar for stimulating discussions about both experimental and data
processing aspects of this work. This work is supporrted by Engineering and Physical Sciences
Research Council (EPSRC) grants EP/W033054/1 and EP/S015973/1. JPF is supported by
EPSRC studentships.

7. Disclosures
The authors declare no conflicts of interest

8. Data availability
Data underlying the results presented may be obtained from the authors upon reason- able request.

References
1. D. M. Mittleman, “Twenty years of terahertz imaging,” Opt. Express 26, 9417–9431 (2018).
2. M. Tonouchi, “Cutting-edge terahertz technology,” Nat. photonics 1, 97–105 (2007).
3. S. Brinkmann, N. Vieweg, G. Gärtner, et al., “Towards quality control in pharmaceutical packaging: Screening
folded boxes for package inserts,” J. Infrared, Millimeter, Terahertz Waves 38, 339–346 (2017).
4. X. Yang, X. Zhao, K. Yang, et al., “Biomedical applications of terahertz spectroscopy and imaging,” Trends
biotechnology 34, 810–824 (2016).
5. A. Leitenstorfer, A. S. Moskalenko, T. Kampfrath, et al., “The 2023 terahertz science and technology roadmap,” J.
Phys. D: Appl. Phys. 56, 223001 (2023).
6. M. G. Gustafsson, “Surpassing the lateral resolution limit by a factor of two using structured illumination microscopy,”
J. microscopy 198, 82–87 (2000).
7. R. Heintzmann and M. G. Gustafsson, “Subdiffraction resolution in continuous samples,” Nat. Photonics 3, 362–364
(2009).
8. X. Chen, S. Zhong, Y. Hou, et al., “Superresolution structured illumination microscopy reconstruction algorithms: a
review,” Light. Sci. & Appl. 12, 172 (2023).
9. Z. Long, T. Wang, C. You, et al., “Terahertz image super-resolution based on a deep convolutional neural network,”
Appl. optics 58, 2731–2735 (2019).
10. Y. Wang, F. Qi, and J. Wang, “Terahertz image super-resolution based on a complex convolutional neural network,”
Opt. letters 46, 3123–3126 (2021).
11. H. Guerboukha, Y. Cao, K. Nallappan, and M. Skorobogatiy, “Super-resolution orthogonal deterministic imaging
technique for terahertz subwavelength microscopy,” ACS Photonics 7, 1866–1875 (2020).
12. C. Wade, N. Šibalić, N. De Melo, et al., “Real-time near-field terahertz imaging with atomic optical fluorescence,”
Nat. Photonics 11, 40 (2017).
13. L. A. Downes, A. R. MacKellar, D. J. Whiting, et al., “Full-field terahertz imaging at kilohertz frame rates using
atomic vapor,” Phys. Rev. X 10, 011027 (2020).
14. R.-W. Lu, B.-Q. Wang, Q.-X. Zhang, and X.-C. Yao, “Super-resolution scanning laser microscopy through virtually
structured detection,” Biomed. optics express 4, 1673–1682 (2013).
15. Y. Lu, T. Son, T.-H. Kim, et al., “Virtually structured detection enables super-resolution ophthalmoscopy of rod and
cone photoreceptors in human retina,” Quant. Imaging Med. Surg. 11, 1060 (2021).
16. F. Wei, D. Lu, H. Shen, et al., “Wide field super-resolution surface imaging through plasmonic structured illumination
microscopy,” Nano letters 14, 4634–4639 (2014).
17. D. Dan, M. Lei, B. Yao, et al., “Dmd-based led-illumination super-resolution and optical sectioning microscopy,” Sci.
reports 3, 1116 (2013).
18. R. Fiolka, M. Beck, and A. Stemmer, “Structured illumination in total internal reflection fluorescence microscopy
using a spatial light modulator,” Opt. Lett. 33, 1629–1631 (2008).
19. Y. Zhi, B. Wang, and X. Yao, “Super-resolution scanning laser microscopy based on virtually structured detection,”
Crit. Rev. Biomed. Eng. 43 (2015).
20. C. S. Adams and I. Hughes, Optics f2f: from Fourier to Fresnel (Oxford University Press, 2019).
21. M. Müller, V. Mönkemöller, S. Hennig, et al., “Open-source image reconstruction of super-resolution structured
illumination microscopy data in imagej,” Nat. communications 7, 10980 (2016).
22. M. Ingaramo, A. G. York, E. Hoogendoorn, et al., “Richardson–lucy deconvolution as a general tool for combining
images with complementary strengths,” ChemPhysChem 15, 794–800 (2014).
23. V. Perez, B.-J. Chang, and E. H. K. Stelzer, “Optimal 2d-sim reconstruction by two filtering steps with richardson-lucy
deconvolution,” Sci. reports 6, 37149 (2016).
24. K. Chu, P. J. McMillan, Z. J. Smith, et al., “Image reconstruction for structured-illumination microscopy with low
signal level,” Opt. express 22, 8687–8702 (2014).
25. A. Lal, C. Shan, and P. Xi, “Structured illumination microscopy image reconstruction algorithm,” IEEE J. Sel. Top.
Quantum Electron. 22, 50–63 (2016).
26. L. A. Downes, L. Torralbo-Campo, and K. J. Weatherill, “A practical guide to terahertz imaging using thermal atomic
vapour,” New J. Phys. 25, 035002 (2023).
27. C. Carr, C. S. Adams, and K. J. Weatherill, “Polarization spectroscopy of an excited state transition,” Opt. Lett. 37,
118–120 (2012).
28. B. E. Jones, J. W. Thomas, A. Selyem, et al., “A compact stabilized three-laser optical pump system for an imaging
system based on THz-to-visible conversion via atomic vapour,” in Quantum Sensing and Nano Electronics and
Photonics XVIII, vol. PC12009 M. Razeghi, G. A. Khodaparast, and M. S. Vitiello, eds., International Society for
Optics and Photonics (SPIE, 2022), p. PC120090J.
29. T. Gallagher, Rydberg Atoms, Cambridge Monographs on Atomic, Molecular and Chemical Physics (Cambridge
University Press, 1994).
30. S. E. Ruzin, Techniques in Light Microscopy (Oxford University Press, 2024).
31. R. Barakat, “Determination of the optical transfer function directly from the edge spread function,” J. Opt. Soc. Am.
55, 1217–1221 (1965).
32. A. Shayei, Z. Kavehvash, and M. Shabany, “Improved-resolution millimeter-wave imaging through structured
illumination,” Appl. Opt. 56, 4454–4465 (2017).