Genuine Resolution: Detection, Recognition & Voting

• Genuine Resolution is an operational concept that distinguishes intrinsic, irreducible distinctions from superficial measurements across various domains.
• In photon-number-resolving detectors, genuine resolution certifies measurement granularity by ensuring outcomes cannot be replicated by reduced, classically simulated measurements.
• In face recognition and voting systems, genuine resolution demands capturing authentic detail and providing unambiguous, evidence-based dispute resolution.

Searching arXiv for the relevant papers and terminology. “Genuine resolution” does not appear in the provided literature as a single uniform technical term. Instead, the literature presents several domain-specific usages centered on a common contrast: between nominal or apparent resolution and an intrinsic, operationally meaningful ability to distinguish, certify, or preserve what is genuinely present. Its clearest formal use is in photon-number-resolving detection, where “genuine resolution” denotes the number of outcomes a detector can irreducibly distinguish in a single shot (Pauwels et al., 12 Jun 2026). Closely related notions include “genuine / underlying resolution” in face recognition, where image dimensions may overstate the facial detail actually present (Wong et al., 2013), and “genuine dispute resolution” in voting, where any third party must be able to determine unambiguously whether a voter or an authority is correct (Basin et al., 2020). Other adjacent uses concern the resolution of mismatches between genuine and spoofed or posed signals in speech and facial-expression analysis (Lapidot et al., 2023, Jia et al., 2020).

1. Distinct technical senses of the term

The provided literature uses the word “genuine” in several non-equivalent ways. In some cases, “resolution” is literal and measurement-theoretic; in others, it concerns latent information content or the public settlement of conflicting claims.

Domain	Usage	Core object
Photonic quantum detection	“genuine resolution”	irreducible measurement-outcome resolution
Face recognition	“genuine / underlying resolution”	actual facial detail rather than image size
Voting security	“genuine dispute resolution”	third-party decidability of disputes
Speech anti-spoofing	“genuinization”	reduction of genuine–spoof PMF mismatch
Facial-expression analysis	genuine vs. posed discrimination	credibility of visible affect

The detector-setting paper gives the most explicit formal definition. A $K$-outcome POVM $\{M_b\}_{b=1}^K$ has genuine resolution of at least $R+1$ if it is not $R$-outcome simulatable, where simulability means reproduction by measurements with at most $R$ outcomes together with shared randomness and classical post-processing (Pauwels et al., 12 Jun 2026). By contrast, the face-recognition paper uses “genuine / underlying resolution” operationally: the true amount of information-bearing facial detail can differ from the stored image size because of downsampling, blur, poor optics, or prior resizing (Wong et al., 2013). The voting paper introduces “genuine dispute resolution” as the property that any third party can unambiguously determine who is right when a voter and an authority make conflicting claims (Basin et al., 2020).

These uses are not identical. A plausible implication is that “genuine resolution” functions as a family resemblance term: it marks cases where surface outputs, nominal dimensions, or public labels are insufficient unless backed by intrinsic discriminative structure, admissible evidence, or content-faithful signal detail.

2. Operational genuine resolution in photon-number-resolving detectors

The most formal treatment appears in the certification framework for photon-number-resolving detectors. A measurement with $K$ outcomes is represented by a POVM $\{M_b\}_{b=1}^K$, with outcome probabilities

$p(b|\rho)=\mathrm{tr}(M_b\,\rho).$

The central question is whether many reported output bins reflect a genuinely fine-grained quantum measurement, or whether a coarser detector is merely followed by classical relabeling. The paper formalizes this by $R$-outcome simulability: $$M_b = \sum_\lambda \sum_{i=1}^R p(\lambda)\, p(b|i,\lambda)\, E_{i|\lambda}.$$ A detector has genuine resolution of at least $\{M_b\}_{b=1}^K$0 if this decomposition is impossible on the relevant subspace $\{M_b\}_{b=1}^K$1 (Pauwels et al., 12 Jun 2026).

This framework makes nominal outcome count insufficient. Even an efficiency-limited idealized PNR detector can lose genuine resolution. For $\{M_b\}_{b=1}^K$2, the detector would ideally distinguish $\{M_b\}_{b=1}^K$3 photons, but the paper shows that when

$\{M_b\}_{b=1}^K$4

the measurement becomes 2-simulatable. Thus three formal outcomes need not imply genuine three-outcome resolution. More generally, the paper determines threshold efficiencies $\{M_b\}_{b=1}^K$5 for different $\{M_b\}_{b=1}^K$6 and $\{M_b\}_{b=1}^K$7 by linear programming over diagonal simulations in the Fock basis (Pauwels et al., 12 Jun 2026).

The certification protocol is prepare-and-measure. Coherent-state probes $\{M_b\}_{b=1}^K$8 are used, and because the detector is not phase-locked, the relevant inputs are phase-averaged coherent states

$\{M_b\}_{b=1}^K$9

The score is the average guessing probability

$R+1$0

If a detector is $R+1$1-outcome simulatable, then in the source-agnostic setting

$R+1$2

Trusted and untrusted source models then refine this with conservative Poisson-tail corrections outside the target subspace $R+1$3 (Pauwels et al., 12 Jun 2026).

Experimentally, the protocol was applied to a 28-pixel photon-number-resolving superconducting nanowire single-photon detector. Using seven coherent probe intensities $R+1$4, the paper reports certification of genuine two-outcome resolution in $R+1$5, genuine three-outcome resolution in $R+1$6, and, as its headline result, genuine four-outcome resolution in $R+1$7 in the trusted setting. For $R+1$8 and $R+1$9, the measured value

$R$0

exceeded the trusted $R$1 witness bound

$R$2

yielding genuine resolution at least 4, whereas the untrusted bound

$R$3

was not violated (Pauwels et al., 12 Jun 2026).

In this setting, “genuine resolution” is therefore an operational benchmark for irreducible measurement granularity. It is invariant under classical post-processing and directly tied to what the underlying POVM can distinguish in one shot.

3. Genuine or underlying resolution in local-feature face recognition

A different but related meaning appears in face recognition under resolution mismatch. The paper argues that image size is not a reliable proxy for the amount of facial detail present. A face stored at $R$4 pixels may in fact contain only the information of $R$5 or $R$6 because of prior downsampling and upsampling, poor optics, defocus, or blur. The key notion is therefore the image’s “underlying resolution”: the resolution level actually supported by the content, regardless of current dimensions (Wong et al., 2013).

The recognition backbone is Multi-Region Histograms (MRH). Each face is partitioned into regions, $R$7 local blocks are normalized to zero mean and unit variance, a 2D DCT is computed, coefficients from the top-left $R$8 part are retained except the $R$9-th term, and each local descriptor is mapped to posterior probabilities over a Gaussian visual dictionary: $R$0 Regional histograms are averaged and compared by $R$1 distance, with optional cohort normalization (Wong et al., 2013).

The paper then builds a dynamic compensation framework with two candidate recognizers: System A at $R$2 and System B at $R$3. System A is best when underlying resolution is approximately $R$4 to $R$5, whereas System B is better for $R$6 and below, especially $R$7 and $R$8. A detector estimates whether each query image is more consistent with a high-resolution class $R$9 or a low-resolution class $K$0 by resizing all inputs to $K$1, extracting MRH features using $K$2 regions and 1024 visual words, and comparing average raw MRH distances

$K$3

The smaller of $K$4 and $K$5 determines which recognition system should be used (Wong et al., 2013).

On a resolution-modified LFW benchmark, the detector achieved about 99% average accuracy in selecting the appropriate system. The dynamic recognizer then obtained 68.09% average accuracy, compared with 64.27% for System A alone and 66.94% for System B alone. The operational content of “genuine resolution” here is thus not a detector’s outcome count but the amount of true spatial detail available for recognition (Wong et al., 2013).

4. Genuine dispute resolution in voting protocols

In voting, the phrase denotes a security and accountability property rather than a signal or measurement property. A dispute arises when a voter claims that the voting authority mishandled a ballot and the authority claims to have followed the protocol. The paper’s criterion is exact: a dispute can be resolved if any third party can unambiguously determine who is right. This goes beyond standard verifiability because individual verifiability failures need not be interpretable by outsiders (Basin et al., 2020).

The paper classifies two central disputes. In D1, the voter claims to have cast ballot $K$6 while the authority claims that $K$7 was not cast. In D2, the voter claims not to have cast ballot $K$8 while the authority claims that $K$9 was cast. These are formalized through trace properties protecting an honest voter and an honest authority. A central innovation is timeliness: by the election’s end, an honest voter must already possess the evidence needed to prove authority misbehavior. The paper encodes public evidence through signals $\{M_b\}_{b=1}^K$0, $\{M_b\}_{b=1}^K$1, and $\{M_b\}_{b=1}^K$2, and requires that any public verdict predicate $\{M_b\}_{b=1}^K$3 depend only on these signals (Basin et al., 2020).

The resulting dispute-resolution framework is packaged as

$\{M_b\}_{b=1}^K$4

with voter-side, authority-side, and functionality conditions. For timeliness, the paper defines $\{M_b\}_{b=1}^K$5 using $\{M_b\}_{b=1}^K$6, $\{M_b\}_{b=1}^K$7, and $\{M_b\}_{b=1}^K$8. Its main theorem is a topology characterization: $\{M_b\}_{b=1}^K$9 Thus timely dispute resolution is possible exactly for topologies at least as strong as one of seven minimal topologies. Reliable communication and either undeniable channels, trusted platforms, or reliable request/reply paths with partial trust assumptions become structurally necessary (Basin et al., 2020).

The case study is MixVote, a mixnet-based protocol. It provides authority-signed confirmations as evidence, proves $p(b|\rho)=\mathrm{tr}(M_b\,\rho).$0, $p(b|\rho)=\mathrm{tr}(M_b\,\rho).$1, $p(b|\rho)=\mathrm{tr}(M_b\,\rho).$2, $p(b|\rho)=\mathrm{tr}(M_b\,\rho).$3, and $p(b|\rho)=\mathrm{tr}(M_b\,\rho).$4, and is also shown to satisfy verifiability and receipt-freeness. Here “genuine resolution” is the public, timely, and symmetric settlement of disputes by admissible evidence rather than by private claims (Basin et al., 2020).

5. Resolving genuine–artificial discrepancies in speech and facial behavior

Two additional literatures use adjacent concepts rather than the exact phrase. In speaker anti-spoofing, the closest notion is “genuinization.” The paper starts from the observation that the waveform PMF of genuine speech differs significantly from the PMF of spoofed speech. It then defines a waveform-domain histogram remapping

$p(b|\rho)=\mathrm{tr}(M_b\,\rho).$5

so that the transformed spoofed waveform approximately satisfies $p(b|\rho)=\mathrm{tr}(M_b\,\rho).$6. Operationally, this is histogram specification from spoofed speech to the genuine reference distribution (Lapidot et al., 2023).

The importance of this transformation is diagnostic. When genuinization is applied to spoofing attacks, spoofing detection performance can degrade by up to a factor of 10. When integrated into countermeasures, it can also improve detection in different cases. The paper therefore treats the genuine–spoof PMF gap as both a vulnerability-analysis tool and a defense-design clue. This is not “genuine resolution” as a named formal term, but it is explicitly the resolution of a genuine–spoof mismatch in raw waveform statistics (Lapidot et al., 2023).

In facial-expression analysis, the review defines the task as credibility assessment: deciding whether a visible expression is genuine or posed. It organizes the literature into muscle movement / Action Unit methods, spatial-pattern methods, visual-feature methods, and hybrid methods. Across datasets and models, the review repeatedly emphasizes temporal structure—neutral $p(b|\rho)=\mathrm{tr}(M_b\,\rho).$7 onset $p(b|\rho)=\mathrm{tr}(M_b\,\rho).$8 apex $p(b|\rho)=\mathrm{tr}(M_b\,\rho).$9 offset $R$0 neutral—and reports that onset-phase features are often especially informative. The field is presented as one of discriminating spontaneous expressions from deliberate or volitional ones rather than merely recognizing emotion category (Jia et al., 2020).

A plausible implication is that these two literatures share with the detector, face-recognition, and voting uses a common methodological demand: the system must distinguish authentic underlying structure from labels or appearances that can be generated without corresponding intrinsic content.

6. Boundaries of the term and related “genuine” constructs

The provided literature also contains several prominent uses of “genuine” that are not usages of “genuine resolution.” These include “GENUINE” as Graph ENhanced mUlti-level uncertaINty Estimation for LLMs (Wang et al., 9 Sep 2025); genuine tripartite steering revealed by sequential measurement (Mukherjee et al., 2017); genuine tripartite steering and entanglement certification in a two-sided device-independent scenario (Gupta et al., 2019); genuine multipartite entanglement as distinct from broader $R$1-entanglement hierarchies (Gerke et al., 2016); the junction law for genuine multi-entropy (Iizuka et al., 18 Feb 2026); genuine and strongly genuine polynomials controlling specialization behavior and Galois-group persistence (Bonolis et al., 2 Jul 2026); and genuine BGK schemes for ultra-relativistic flow simulation, where “genuine” refers to inclusion of actual model collisions in the gas transport process (Chen et al., 2017).

These are distinct technical vocabularies. They concern irreducible multipartite quantum correlations, higher-order entanglement diagnostics, algebraic non-reducibility under specialization, structure-aware uncertainty pooling, or collision-faithful kinetic flux construction. They do not define a general theory of “genuine resolution” across disciplines.

What they do suggest, however, is a recurring technical role for the adjective “genuine.” It often marks a refusal to equate surface form with intrinsic capability: a detector with many bins may not genuinely resolve many photon numbers; a $R$2 face image may not genuinely contain $R$3 facial detail; a voting protocol may detect anomalies without genuinely resolving disputes; and a spoofed waveform or posed expression may imitate surface appearance without sharing the statistics or dynamics of genuine signals. In the provided literature, “genuine resolution” is therefore best understood not as a single doctrine, but as a family of operational criteria for distinguishing irreducible, content-faithful, or publicly certifiable structure from nominal or post-processed appearance.