Pith. sign in

REVIEW 4 major objections 5 minor 25 references

Passive acoustic non-line-of-sight localization without a relay surface

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Diffracted sound from a doorway edge can localize a hidden hand clap in 3D, and a spectral-ratio trick recovers azimuth around a single corner.

desk verdict A genuinely new passive-acoustic NLOS localization idea with a solid doorway proof-of-concept, but the corner/azimuth claim is qualitative and one key parameter looks like a typo. read the letter →

arxiv 2506.08471 v1 pith:76Q5YORS submitted 2025-06-10 cs.SD eess.ASeess.SP

Tal I. Sommer , Ori Katz This is my paper
classification cs.SDeess.ASeess.SP
keywords non-line-of-sightlocalizationpassiveacousticsknife-edgediffractionlossspectralratiotime-of-arrivalbeamforminghead-relatedtransferfunction
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper claims that a hidden acoustic source can be localized in three dimensions using only sound that diffracts around the edge of an obstacle, with no reflecting relay surface such as a wall or floor. In a doorway scene, the two door edges act as virtual line arrays: fitting the measured arrival times of the first and second wavefronts to a spherical-wave model recovers range, azimuth, and height, with reported RMSEs near 18%, 3 degrees, and 6%. In a single-edge corner scene, the same timing fit gives range and height, while azimuth is recovered from the ratio of diffracted spectra measured by two arrays at different angles, which cancels the unknown source spectrum. The paper presents this as a proof of principle that edge diffraction, not just reflection, can carry enough information for non-line-of-sight acoustic localization in environments where relay surfaces are absent.

What carries the argument

Knife-edge diffraction is the carrier: a sharp edge bends incident sound toward the detector with an amplitude loss given by the Fresnel-Kirchhoff integral $L(\nu) = \frac{1+j}{2}\int_\nu^\infty \exp(-j\pi t^2/2)\,dt$, where $\nu \propto \theta\sqrt{f}$. This identity does double duty. Its phase and arrival structure make the edge act as a virtual point of emission whose wavefront curvature and time of arrival encode range and height through $TOA(z) = t_0 + \sqrt{(z-z_0)^2 + R_0^2}$; its frequency-dependent magnitude provides the azimuth cue, because lower frequencies diffract more efficiently at large angles. The paper turns door edges into a two-edge virtual relay surface for beamforming, and turns two arrays at different azimuths into an unknown-spectrum-cancelling spectral-ratio estimator, the acoustic analogue of the head-related transfer function.

What would settle it

In the single-edge corner setup, sweep the spectral-analysis window from roughly 10 microseconds to 10 milliseconds on the same hand-clap recordings: if the spectral-ratio curves for 5, 10, and 15 degree azimuths overlap for all window sizes, the reported azimuth discrimination is an artefact of the chosen window rather than a property of knife-edge diffraction. Alternatively, deliberately perturb the measured array-to-edge distances in the doorway scene by 10 cm and check whether the azimuth estimate shifts by more than the claimed 3 degree RMSE when no correction is applied.

Watch

Extended reading notes

Core claim

The central claim is that edge diffraction from a visible doorway or building corner is sufficient for passive 3D localization of a hidden point source. For a doorway, the two vertical edges form two virtual line arrays; the spherical time-of-arrival relation $TOA(z) = t_0 + \sqrt{(z-z_0)^2 + R_0^2}$ fitted to the first arriving diffracted wavefront gives propagation distance and source height, and a delay-and-sum-like amplitude metric using both edges locates the source in the horizontal plane. For a single convex corner, the first-arriving diffracted wavefront likewise gives range and height, and azimuth is read from the spectral ratio $L_\theta(f)/L_{\theta+\Delta\theta}(f)$ measured by two arrays separated by 25 degrees, which removes dependence on the unknown emission spectrum, much as the head-related transfer function lets humans judge source directions from spectral cues. The paper reports 18% range RMSE, 3 degree azimuth RMSE, and 6% height RMSE in the doorway experiment, and demonstrates corner azimuth estimates at source azimuths of 5, 10, and 15 degrees, matched against FDTD simulations. The quoted accuracies are obtained after small corrections of 2 to 8 cm in the measured array-to-edge distances, within the stated measurement uncertainty.

Load-bearing premise

The localization accuracy rests on knowing the visible geometry, especially the distance between each microphone array and the diffracting edge, to within a few centimetres, and on choosing the spectral-analysis window so that the ratio separates azimuths; if either must be hand-tuned per scene, the method is not yet a blind measurement.

Editorial extensions

If this is right

  • Passive NLOS localization becomes possible where no reflective relay surface exists, including open doorways and building corners, the two geometries demonstrated.
  • Source azimuth remains recoverable without knowing what the source sounds like, because the two-array spectral ratio cancels the unknown emitted spectrum.
  • A single vertical microphone array can already recover range and height from the curvature and timing of the diffracted wavefront; the second array is needed only for the final angular coordinate.
  • The method works with broadband impulsive sounds such as hand claps in reverberant concrete environments, suggesting that ordinary everyday sounds may serve as stimuli.
  • Accuracy is bounded by how well the visible scene is known, since centimetre-level errors in array-to-edge distances translate directly into azimuth bias; precise geometric metrology is part of the method.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Because the diffraction parameter satisfies $\nu \propto \theta\sqrt{f}$, a broadband measurement folds many independent frequency constraints into one azimuth estimate; this may allow a single array to estimate azimuth when the source spectrum is known or separately measured, and would make the estimate less sensitive to narrowband ambient noise.
  • The same two-array spectral ratio could be used as a calibration tool: with a source at a known position, it could measure the effective scene geometry and window response, potentially removing the need for the manual centimetre-level corrections the paper applies.
  • The temporal window for the corner azimuth is written as 0.7 microseconds, far shorter than the roughly 0.2 ms period of the 5 kHz hand-clap content; if that value is literal, the spectral ratio is selecting a very brief slice of the first wavefront, and the method may not transfer to longer or non-impulsive emissions without different processing.
  • The two-array spectral-ratio estimator is essentially a binaural, HRTF-inspired cue; a natural next test is whether the ratio can track a moving source around a corner, or whether replacing one array with a direction-dependent filter yields the same azimuth discrimination.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper proposes passive acoustic non-line-of-sight (NLOS) localization using edge diffraction rather than a visible relay surface. In the doorway scenario, a vertical microphone array measures the diffracted wavefront from the near door edge and the reflected wavefront from the far edge; the time-of-arrival curvature gives range and height, and a modified beamforming metric over the two virtual line-arrays gives azimuth. Reported experimental RMSEs are 18% in range, 3° in azimuth, and 6% in height across 15 single hand-clap trials. In the single-edge/convex-corner scenario, range and height are obtained from the first-arriving diffracted wavefront, while azimuth is estimated from the spectral ratio of signals at two arrays separated by 25°, exploiting the frequency dependence of knife-edge diffraction. The paper reports one-standard-deviation bands for source azimuths of 5°, 10°, and 15° compared with FDTD simulations, but does not provide a quantitative inversion or azimuth errors. The forward model is cross-checked with k-Wave FDTD simulations.

Significance. The idea of using knife-edge diffraction for passive acoustic NLOS localization is novel and potentially useful, especially for scenarios where no relay surface is available. Strengths: the physical forward model is standard and independently cross-checked with FDTD; the experiments use real impulsive sources and multiple trials; and the manuscript is candid about sensitivities to scene geometry and temporal window size. However, the validation as presented is incomplete: the single-edge azimuth estimate is essentially a visual comparison, and the doorway accuracy numbers are conditional on post hoc geometry corrections and an empirically tuned metric. If the proposed inversion and blind-validation steps are added, the result would be a meaningful proof of principle; in its current form the central claim of '3D localization without a relay surface' is not yet fully supported.

major comments (4)
  1. [Single-edge/convex-corner scenario, Fig. 5c] The azimuth-retrieval step, which is the load-bearing element for the claim of localization without a relay surface, is presented only as a visual comparison. Fig. 5c overlays one-standard-deviation bands for three source azimuths on FDTD curves, but the paper does not define an inversion from a measured spectral ratio to an azimuth estimate, report per-trial classification, give an azimuth RMSE, or perform a blind or held-out test. The text also states that invalid measurements were disregarded, but the exclusion criteria (low intensity, undetectable wavefront, fit RMSE larger than 0.1 ms) are not reported as counts or applied in a way that would allow the reader to judge selection bias. As written, this supports separation of the three tested azimuths in the ensemble, not a demonstrated localization capability. Please provide a concrete estimator (for example, a least-squares fit to the FDTD ratio library or a classifier), per-trial errors, and a validation protocol that does not select on the outcome.
  2. [Convex-corner experiment, temporal window paragraph] The spectral-ratio result is reported after fine-tuning a temporal window until optimal results were produced, at a stated size of 0.7 usec. A 0.7 μs window cannot resolve the 500 Hz–9 kHz band (the frequency resolution would be about 1.4 MHz), so this is either a typo for a much larger window or an unreproducible processing choice that the paper itself concedes is sensitive. Please correct the units and, more importantly, fix the window-selection rule (for example, a fixed number of wavefront periods or a data-driven criterion) and report the sensitivity of the spectral ratio to the window length. Without this, the corner azimuth result is not reproducible by an independent group.
  3. [Doorway experiment, Eq. (1), Fig. 2(c-e)] The quoted RMSEs (range about 18%, azimuth about 3°, height about 6%) are obtained only after applying post hoc corrections of 2–8 cm to the measured array-to-edge distances, described as within experimental measurement error. Because the azimuth bias is stated to be sensitive to these distances, the accuracy claims are conditional on unquantified scene-geometry calibration. Please report the uncorrected RMSEs, provide an independent uncertainty estimate for Rd2 and Rr2, and show how the azimuth error scales with these parameters (for example, via the analytic derivative of Eq. (1) or a sensitivity scan). If the corrections are necessary, they should be specified as a calibration procedure applicable to unseen scenes (for example, measured positions before the experiment) rather than as per-run adjustments.
  4. [Doorway reconstruction metric, Eq. (2)] The reconstruction metric M uses an empirically chosen weighting (1/RdRr)^2, which the text says was found to give better results than conventional delay-and-sum. Since the final RMSEs depend on this choice, please report a comparison with the conventional delay-and-sum implementation on the same dataset and a sensitivity test over the weighting exponent. Otherwise the quoted accuracies are tied to an unstated tuning step and cannot be assessed as a property of the proposed method.
minor comments (5)
  1. [Throughout] There are several typos and grammatical errors, including 'surfac e' in the title, 'rely surface' instead of 'relay surface', 'emmissions' instead of 'emissions', and 'utilize' agreement issues. A careful proofread is needed.
  2. [Convex-corner experiment, text near Fig. 5] The text says 'Fig. 5b compares the experimentally measured spectral ratios,' but Fig. 5b is the amplitude envelope; the comparison to FDTD spectral ratios is in Fig. 5c. Please correct the cross-reference.
  3. [Convex-corner experiment, Supplementary Fig. S1] The reference to the hand-clap spectrum appears as 'Supplementary Fig. ??' in the corner scenario section; the figure number should be filled in.
  4. [Eq. (2)] Equation (2) appears to have a missing operator between the Ad and Ar terms; as printed, 'M(Rd,Rr)=Ad/(Rd)^2 Ar/(Rr)^2' is not a well-defined expression. Please clarify whether the intended metric is a product, a sum, or a ratio of the two amplitude terms.
  5. [Data availability] The data availability statement says data are available 'upon reasonable request.' For a results-driven paper that depends on several processing choices, depositing the raw microphone recordings and processing scripts in a public repository would strengthen reproducibility.

Circularity Check

2 steps flagged · score 4.0 of 10

Central TOA/Fresnel derivation is independent, but the headline validation numbers are partly in-sample: scene geometry is corrected by 2–8 cm before quoting azimuth RMSE, and the corner spectral ratio relies on a fine-tuned temporal window without a defined azimuth inversion.

  1. fitted input called prediction [Doorway experiment, paragraph beginning 'Note that the reconstructed location is sensitive...' (after Fig. 2c-e)]
    "Note that the reconstructed location is sensitive to errors in the estimation/measurement of the geometry of the visible part of the scene. In particular, the measured distance between the array and the edges (Rd2 and Rr2) causes an additive constant bias to the azimuth. The presented results in Fig.2b are obtained after a small change of 2 cm − 8cm in the measured scene parameters, which is within our experimental measurement error."

    The scene constants Rd2 and Rr2 are fixed inputs to the TOA model (Eq. 1) and to the beamforming metric. The reported azimuth RMSE (about 3 degrees) is quoted only after these inputs were changed by 2–8 cm to remove the azimuth bias on the same experimental dataset. As written, no independent re-measurement of the corrected constants is reported, so the accuracy figure is not an out-of-sample prediction from the measured geometry: the input has been calibrated using the same source positions whose errors are then reported. This is not a per-trial fit, but it folds a post hoc bias correction into the headline localization accuracy.

  2. fitted input called prediction [Single-edge/corner scenario, paragraph beginning 'To optimally filter out background noises...' (see also Fig. 5c)]
    "We found that the spectral-ratio is sensitive to the exact choice of window size. In our experiments the window size was fine-tuned until optimal results were produced, at a window-size of 0.7usec."

    In the no-relay-surface corner scenario, the only evidence for azimuth retrieval is the spectral-ratio separation shown in Fig. 5c. The temporal window used to compute those spectra was selected post hoc 'until optimal results were produced', and measurements with poor wavefront fits or low intensity were discarded afterwards. The paper does not specify an inversion from measured spectral ratio to azimuth, nor does it report per-trial azimuth errors; the displayed agreement between experimental bands and FDTD curves is therefore a selected outcome rather than a blind prediction. The stated 0.7 microsecond window is also inconsistent with resolving the 0.5–9 kHz band, so the processing step is not reproducible as written.

full rationale

The paper's forward derivations are standard and self-contained: Eq. 3 is the classical Fresnel-Kirchhoff knife-edge diffraction loss, and Eq. 1 is the spherical time-of-arrival model. Both are cross-checked against k-Wave FDTD simulations with explicit scene parameters, so there is no equation-level self-definition and no load-bearing self-citation. The doorway range and height estimates follow directly from the TOA wavefront curvature and retain independent content. The circularity is confined to the validation pipeline: the quoted azimuth performance is obtained after a 2–8 cm correction to the scene geometry, and the corner spectral-ratio result is obtained after fine-tuning the processing window and excluding trials, with no defined inversion from ratio to azimuth. These steps mean the reported headline numbers are not fully out-of-sample, but they do not reduce the underlying diffraction/TOA derivation to its inputs. A score of 4 reflects this partial, post hoc calibration rather than constructional circularity.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

The central claim rests on standard diffraction theory and a spherical-wave TOA model, plus assumptions about time-gating and scene geometry accuracy. No new physical entities are introduced. The main free parameters are post hoc corrections and an empirically chosen weighting/window, which limit the strength of the validation.

free parameters (3)
  • Doorway scene-geometry correction = 2-8 cm shift in measured array-to-edge distances Rd2, Rr2
    Applied after inspecting doorway localization results to remove an additive azimuth bias; claimed to be within experimental measurement error.
  • Spectral-ratio temporal window size = 0.7 μs as written (likely 0.7 ms typo)
    Window length was fine-tuned until optimal spectral-ratio results were produced; the azimuth discrimination in Fig. 5c depends on it.
  • Beamforming metric weighting factor = (1/(Rd Rr))^2
    Empirically chosen weighting in the localization metric M to prioritize close locations and suppress late reverberations.
assumptions (6)
  • domain assumption Fresnel-Kirchhoff knife-edge diffraction loss L(ν) with ν ∝ θ√f describes the amplitude of a diffracted wave (Eq. 3).
    Standard theory from scalar diffraction, cited to reference 15; used to motivate spectral-ratio azimuth estimation.
  • domain assumption The time of arrival of the diffracted or reflected wavefront across the vertical array follows TOA(z) = t0 + sqrt((z-z0)^2 + R0^2) (Eq. 1).
    Assumes the edge acts as a point-like virtual source for the first arrival and that the path length equals a straight line from an image source; not formally derived for the doorway geometry.
  • domain assumption The spectral ratio of signals measured at two arrays equals the ratio of diffraction losses Lθ(f)/Lθ+Δθ(f), canceling the unknown source spectrum.
    Requires identical source emission and negligible path-dependent filtering other than diffraction within the gated window; the authors note extra modulations attributed to reflections from the human body.
  • domain assumption Reverberations can be excluded by time-gating the first arriving wavefront around its TOA.
    Used in both processing pipelines; assumes the early part of the response in a concrete basement is dominated by the direct diffracted/reflected wave.
  • domain assumption The visible scene geometry (array-to-edge distances and angles) is known accurately enough that cm-level errors only cause small bias.
    Load-bearing for TOA fitting; the authors report correcting measured parameters by 2-8 cm after evaluating results, indicating the assumption was not initially met.
  • standard math k-Wave FDTD simulation with linear adiabatic equation of state and highly reflective walls reproduces the relevant wave physics.
    Toolbox (refs 19, 20) is established; simulations used to generate spectral-ratio curves for comparison in Fig. 5c.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Passive acoustic non-line-of-sight localization without a relay surface." pith.science (2026). https://pith.science/paper/76Q5YORS

@misc{pith2026250608471,
  author       = {Pith},
  title        = {Pith review of: Passive acoustic non-line-of-sight localization without a relay surface},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/76Q5YORS}},
  note         = {Machine review of arXiv:2506.08471}
}
read the original abstract

The detection and localization of a source hidden outside the Line-of-Sight (LOS) traditionally rely on the acquisition of indirect signals, such as those reflected from visible relay surfaces such as floors or walls. These reflected signals are then utilized to reconstruct the obscured scene. In this study, we present an approach that utilize signals diffracted from an edge of an obstacle to achieve three-dimensional (3D) localization of an acoustic point source situated outside the LOS. We address two scenarios - a doorway and a convex corner - and propose a localization method for each of them. For the first scenario, we utilize the two edges of the door as virtual detector arrays. For the second scenario, we exploit the spectral signature of a knife-edge diffraction, inspired by the human perception of sound location by the head-related transfer function (HRTF). In both methods, knife-edge diffraction is utilized to extend the capabilities of non-line-of-sight (NLOS) acoustic sensing, enabling localization in environments where conventional relay-surface based approaches may be limited.

Figures

Figures reproduced from arXiv: 2506.08471 by the authors.

Figure 1
Figure 1. FIG. 1. (a) Conceptual illustration of NLOS localization in a [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. (a) The amplitude envelope of the signal detected by t [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3. Conceptual illustration of NLOS localization in a [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: FIG. 4. Theoretical knife-edge diffraction loss for a simplistic [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: FIG. 5. (a,b) Experimental amplitude envelope of the signals [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

25 extracted references · 20 canonical work pages

  1. [1]

    merlin.mbs aapmrev4-1.bst 2010-07-25 4.21a (PWD, AO, DPC) hacked

    FUNCTION id.bst "merlin.mbs aapmrev4-1.bst 2010-07-25 4.21a (PWD, AO, DPC) hacked" ENTRY address archive archivePrefix author bookaddress booktitle chapter collaboration doi edition editor eid eprint howpublished institution isbn issn journal key language month note number organization pages primaryClass publisher school SLACcitation series title translat...

    2010

  2. [2]

    merlin.mbs aipauth4-1.bst 2010-07-25 4.21a (PWD, AO, DPC) hacked

    FUNCTION id.bst "merlin.mbs aipauth4-1.bst 2010-07-25 4.21a (PWD, AO, DPC) hacked" ENTRY address archive archivePrefix author bookaddress booktitle chapter collaboration doi edition editor eid eprint howpublished institution isbn issn journal key language month note number organization pages primaryClass publisher school SLACcitation series title translat...

    2010

  3. [3]

    merlin.mbs aipnum4-1.bst 2010-07-25 4.21a (PWD, AO, DPC) hacked

    FUNCTION id.bst "merlin.mbs aipnum4-1.bst 2010-07-25 4.21a (PWD, AO, DPC) hacked" ENTRY address archive archivePrefix author bookaddress booktitle chapter collaboration doi edition editor eid eprint howpublished institution isbn issn journal key language month note number organization pages primaryClass publisher school SLACcitation series title translati...

    2010

  4. [4]

    merlin.mbs apsrev4-1.bst 2010-07-25 4.21a (PWD, AO, DPC) hacked

    FUNCTION id.bst "merlin.mbs apsrev4-1.bst 2010-07-25 4.21a (PWD, AO, DPC) hacked" ENTRY address archive archivePrefix author bookaddress booktitle chapter collaboration doi edition editor eid eprint howpublished institution isbn issn journal key language month note number organization pages primaryClass publisher school SLACcitation series title translati...

    2010

  5. [5]

    merlin.mbs apsrmp4-1.bst 2010-07-25 4.21a (PWD, AO, DPC) hacked

    FUNCTION id.bst "merlin.mbs apsrmp4-1.bst 2010-07-25 4.21a (PWD, AO, DPC) hacked" ENTRY address archive archivePrefix author bookaddress booktitle chapter collaboration doi edition editor eid eprint howpublished institution isbn issn journal key language month note number organization pages primaryClass publisher school SLACcitation series title translati...

    2010

  6. [6]

    Faccio , author A

    author author D. Faccio , author A. Velten , \ and\ author G. Wetzstein ,\ title title Non-line-of-sight imaging , \ @noop journal journal Nature Reviews Physics \ volume 2 ,\ pages 318--327 ( year 2020 ) NoStop

    work page 2020

  7. [7]

    Velten , author T

    author author A. Velten , author T. Willwacher , author O. Gupta , author A. Veeraraghavan , author M. G. \ Bawendi , \ and\ author R. Raskar ,\ title title Recovering three-dimensional shape around a corner using ultrafast time-of-flight imaging , \ @noop journal journal Nature communications \ volume 3 ,\ pages 745 ( year 2012 ) NoStop

    work page 2012

  8. [8]

    Gariepy , author F

    author author G. Gariepy , author F. Tonolini , author R. Henderson , author J. Leach , \ and\ author D. Faccio ,\ title title Detection and tracking of moving objects hidden from view , \ @noop journal journal Nature Photonics \ volume 10 ,\ pages 23--26 ( year 2016 ) NoStop

    work page 2016

Show all 25 references
  1. [9]

    O’Toole , author D

    author author M. O’Toole , author D. B. \ Lindell , \ and\ author G. Wetzstein ,\ title title Confocal non-line-of-sight imaging based on the light-cone transform , \ @noop journal journal Nature \ volume 555 ,\ pages 338--341 ( year 2018 ) NoStop

    2018

  2. [10]

    author author D. B. \ Lindell , author G. Wetzstein , \ and\ author V. Koltun ,\ title title Acoustic non-line-of-sight imaging , \ in\ @noop booktitle Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition \ ( year 2019 )\ pp.\ pages 6780--6789 NoStop

    2019

  3. [11]

    Czajkowski \ and\ author J

    author author R. Czajkowski \ and\ author J. Murray-Bruce ,\ title title Two-edge-resolved three-dimensional non-line-of-sight imaging with an ordinary camera , \ @noop journal journal Nature Communications \ volume 15 ,\ pages 1162 ( year 2024 ) NoStop

    2024

  4. [12]

    author author K. L. \ Bouman , author V. Ye , author A. B. \ Yedidia , author F. Durand , author G. W. \ Wornell , author A. Torralba , \ and\ author W. T. \ Freeman ,\ title title Turning corners into cameras: Principles and methods , \ in\ @noop booktitle Proceedings of the ...

    2017

  5. [13]

    author author S. W. \ Seidel , author J. Murray-Bruce , author Y. Ma , author C. Yu , author W. T. \ Freeman , \ and\ author V. K. \ Goyal ,\ title title Two-dimensional non-line-of-sight scene estimation from a single edge occluder , \ @noop journal journal IEEE Transactions ...

    2020

  6. [14]

    Boger-Lombard , author Y

    author author J. Boger-Lombard , author Y. Slobodkin , \ and\ author O. Katz ,\ title title Non-line-of-sight passive acoustic localization around corners , \ @noop journal journal arXiv preprint arXiv:2211.03453 \ ( year 2022 ) NoStop

    2022 arXiv

  7. [15]

    Boger-Lombard \ and\ author O

    author author J. Boger-Lombard \ and\ author O. Katz ,\ title title Passive optical time-of-flight for non line-of-sight localization , \ @noop journal journal Nature communications \ volume 10 ,\ pages 3343 ( year 2019 ) NoStop

    2019

  8. [16]

    Katz , author P

    author author O. Katz , author P. Heidmann , author M. Fink , \ and\ author S. Gigan ,\ title title Non-invasive single-shot imaging through scattering layers and around corners via speckle correlations , \ @noop journal journal Nature photonics \ volume 8 ,\ pages 784--790 ( ...

    2014

  9. [17]

    Katz , author E

    author author O. Katz , author E. Small , \ and\ author Y. Silberberg ,\ title title Looking around corners and through thin turbid layers in real time with scattered incoherent light , \ @noop journal journal Nature photonics \ volume 6 ,\ pages 549--553 ( year 2012 ) NoStop

    2012

  10. [18]

    Sultan , author S

    author author T. Sultan , author S. A. \ Reza , \ and\ author A. Velten ,\ title title Towards a more accurate light transport model for non-line-of-sight imaging , \ @noop journal journal Optics Express \ volume 32 ,\ pages 7731--7761 ( year 2024 ) NoStop

    2024

  11. [19]

    Liu , author S

    author author X. Liu , author S. Bauer , \ and\ author A. Velten ,\ title title Phasor field diffraction based reconstruction for fast non-line-of-sight imaging systems , \ @noop journal journal Nature communications \ volume 11 ,\ pages 1645 ( year 2020 ) NoStop

    2020

  12. [20]

    author author S. V. \ Hum ,\ title title Radio and microwave wireless systems , \ @noop journal journal University of Toronto \ ( year 2007 ) NoStop

    2007

  13. [21]

    author author E. M. \ Wenzel , author M. Arruda , author D. J. \ Kistler , \ and\ author F. L. \ Wightman ,\ title title Localization using nonindividualized head-related transfer functions , \ @noop journal journal The Journal of the Acoustical Society of America \ volume 94 ...

    1993

  14. [22]

    author author C. I. \ Cheng \ and\ author G. H. \ Wakefield ,\ title title Introduction to head-related transfer functions (hrtfs): Representations of hrtfs in time, frequency, and space , \ @noop journal journal journal of the Audio Engineering Society \ volume 49 ,\ pages 23...

    2001

  15. [23]

    Bruschi , author L

    author author V. Bruschi , author L. Grossi , author N. A. \ Dourou , author A. Quattrini , author A. Vancheri , author T. Leidi , \ and\ author S. Cecchi ,\ title title A review on head-related transfer function generation for spatial audio , \ @noop journal journal Applied S...

    2024

  16. [24]

    author author B. E. \ Treeby \ and\ author B. T. \ Cox ,\ title title k-wave: Matlab toolbox for the simulation and reconstruction of photoacoustic wave fields , \ @noop journal journal Journal of biomedical optics \ volume 15 ,\ pages 021314--021314 ( year 2010 ) NoStop

    2010

  17. [25]

    author author B. E. \ Treeby , author J. Budisky , author E. S. \ Wise , author J. Jaros , \ and\ author B. Cox ,\ title title Rapid calculation of acoustic fields from arbitrary continuous-wave sources , \ @noop journal journal The Journal of the Acoustical Society of America...

    2018

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.