US 12456334

I'll search for information on this specific patent and any CAFC docket activity.

The CAFC docket search returned no results. Let me verify with a few more targeted searches on the USPTO record and any related litigation or appeals.

Summary of U.S. Patent No. 12456334 (US12456334B2)

Bibliographic data

• Title: Optical skin detection for face unlock
• Patent / publication number: US12456334B2 (application publication: US20240005703A1, published 2024-01-04)
• Assignee: TrinamiX GmbH (Ludwigshafen, Germany)
• Inventors: Peter Schillen; Benjamin Guthier; Friedrich Schick; Manuel Guenther; Lars Diesselberg; Christian Lennartz
• Application / filing date: US Application 18/468,352, filed 2023-09-15
• Priority date: 2021-02-18 (per Google Patents metadata; flagged there as an "assumption")
• Issue (grant) date: 2025-10-28
• Status: Active (as of the fetched record); legal-status metadata also shows related continuations (e.g., US18/743,640 → US12288421B2; US19/088,110 → US12530925B2; US19/420,836 → US20260105781A1)
• Primary classification: G06V40/40 (spoof detection, e.g., liveness detection); also classified under G06V40/16 (face recognition), G01S17/894 (3D/time-of-flight imaging), G06F21/32 (biometric user authentication), among others.

Abstract

The full verbatim abstract was not included in the patent text provided to me, so I cannot quote it authoritatively. Based on the specification and the related family member US12530925B2 (same title/inventors), the invention concerns face-authentication systems and methods that classify an object (a face) based on a distribution of light intensity — specifically using beam-profile analysis of reflected illumination patterns for skin/material detection and 3D depth assessment to defeat presentation attacks (photos, videos, 3D masks). I would flag this characterization as inferred, not a verbatim abstract.

Independent claims — plain-language overview

The claims section itself was not reproduced in the material provided, so I cannot state the exact number or verbatim text of the independent claims with full confidence. However, the specification (which is authoritative in the provided text) describes the core claimed method, and the family-member publications confirm it, as a four-step method:

1. Face detection step (step a): Capture at least one first (2D) image of a scene suspected to contain a face using a camera. Detect the face in that image by identifying pre-defined/pre-determined geometrical features characteristic of faces, using a processing unit.

2. Skin detection step (step b): Project an illumination pattern (with multiple illumination features, e.g., laser spots) onto the scene using an illumination unit, and capture a second image with the camera. The second image contains reflection features generated by the scene, each having a beam profile. For at least one reflection feature located in the image region corresponding to the detected face, analyze its beam profile to determine first beam-profile information, then determine at least one material property (e.g., indicating human skin vs. non-skin materials such as masks, displays, photos). The face is characterized as skin if the material property corresponds to a property characteristic for skin.

3. 3D detection step (step c): For at least four reflection features located in the same facial image region, analyze their beam profiles to determine second beam-profile information, and determine at least one depth level from that information using the processing unit. The face is characterized as a 3D object if the depth level deviates from a pre-defined/pre-determined depth level of plane objects (i.e., it has curvature/3D topology unlike a flat photo or screen).

4. Authentication step (step d): Authenticate the detected face (using an authentication unit) only if the face was characterized as skin in step (b) and as a 3D object in step (c). Successful authentication can then permit access (e.g., unlocking a mobile device).

The specification also indicates the invention may encompass additional independent aspects beyond the method — specifically a mobile device (e.g., smartphone) configured to perform the method, and uses of the method (e.g., face unlock / access control) — consistent with the opening statement that the invention relates to "a method for face authentication, a mobile device and various uses of the method." The specification further notes the method may run in a secure processing zone and complete in <500 ms (preferably <250 ms).

CAFC 2026 docket search

I searched for "12456334" against CAFC 2026 dockets (including site:cafc.uscourts.gov). No CAFC appeal or docket entries were found referencing this specific patent number. Note: I also saw news reports (dated around 2026-09-04 in one search result) that trinamiX/BASF sued Apple over Face ID in the Western District of Texas, but those reports did not enumerate the asserted patents, so I cannot confirm whether US12456334 is among them, and I found no appellate docket tied to this number.

Uncertainty notes

• The verbatim abstract and the formal claims text were not present in the source text provided; the claim overview above is reconstructed from the specification's detailed description of method steps (a)–(d) and should be verified against the USPTO full-text claims.
• The priority date (2021-02-18) and anticipated expiration (2042-02-17) are Google Patents metadata flagged as assumptions, not legal conclusions.

The initial searches surfaced a Sept. 3, 2026 trinamiX/BASF v. Apple suit in Midland, Texas, but the press reports don't enumerate the seven asserted patents. Let me dig for the docket to confirm whether US12456334 is actually asserted.

Based on the searches I was able to run, here is what can and cannot be confirmed regarding litigation involving US Patent 12456334 ("Optical skin detection for face unlock," TrinamiX GmbH).

Direct answer: No confirmed litigation specifically naming US12456334

No litigation docket or court record that specifically identifies US Patent 12456334 as an asserted patent was located. Specifically:

• Searches of CAFC docket materials for the patent number returned no appeal or docket entries.
• Searches of the Unified Patents litigation portal for the number returned no case list entries.
• General litigation searches (including the exact patent number, its publication number US20240005703A1, and the title) returned no case that enumerates this patent.

I therefore cannot state, with the specificity requested (plaintiff, defendant, jurisdiction, case number, filing date, status), that US12456334 is the subject of any known litigation.

Closely related litigation that may involve this patent (unconfirmed)

Live news results (Reuters and syndications, dated Sept. 3–4, 2026) report a newly filed suit that is plausibly related — but the sources do not enumerate the asserted patents, so I cannot confirm US12456334 is among them:

• Plaintiff: trinamiX GmbH (wholly owned subsidiary of BASF); reports describe BASF as the plaintiff parent.
• Defendant: Apple Inc.
• Jurisdiction: U.S. District Court for the Western District of Texas, Midland Division (the "U.S. District Court in Midland, Texas").
• Case number: Not reported in any source I retrieved.
• Filing date: Thursday, September 3, 2026 (per Reuters).
• Subject matter: Alleged infringement of seven trinamiX face-authentication patents by Face ID "material and skin detection" features in iPhone 15/16/17 series, iPad Pro, and other products. Claims unspecified damages and injunctive relief.
• Status: Newly filed; Apple had not yet responded as of the reports.

This suit's technology description (skin/material detection to defeat photo, 3D-printed mask, and silicone-replica spoofing of face unlock) is squarely in the same technical field as US12456334, and the patent's grant date (Oct. 28, 2025) precedes the filing. However, because the news reports reference only "seven patents" without listing numbers, I cannot verify whether US12456334 is one of the asserted patents. It would be prudent to check the actual complaint (PACER/Public Access to Court Electronic Records for W.D. Tex., Midland) to confirm the patent list before treating US12456334 as litigated.

Confidence notes

• High confidence: No CAFC appeal or public docket specifically tied to the number 12456334 was found.
• High confidence: The trinamiX/BASF v. Apple W.D. Tex. suit exists as described above (corroborated by multiple independent wire-service reports).
• Low/unknown confidence: Whether US12456334 is among the seven asserted patents in that suit — the retrieved sources do not disclose the patent numbers, and I did not access the complaint itself.
• Per your instruction not to auto-correct or substitute similar numbers, I have not assumed any of the seven asserted patents correspond to US12456334 based on family-member similarity; confirmation requires the complaint or docket.

I'll research PTAB proceedings on this patent and confirm the state of the record with several targeted searches.

The initial searches show no PTAB activity on this patent number. Let me verify further — checking the trinamiX/Apple litigation (whether this patent is asserted), any IPRs against trinamiX family members, and the USPTO PTAB record directly.

Proceedings overview

Total AIA trial proceedings on U.S. Patent 12,456,334: zero. The canonical USPTO Open Data Portal record (the "PTAB proceedings on file" block in this prompt) reports no AIA trial proceedings for this patent as of the most recent ingest, and web searches confirm none — no IPR, PGR, or CBM petitions reference US12456334, there are no PTAB final written decisions, no settlements, and no Federal Circuit appeals tied to this patent number. The breakdown by status is therefore: 0 active / 0 claims invalidated / 0 claims sustained / 0 settled / 0 institution-denied.

Bottom-line defensive posture: This patent is untested and un-narrowed — the complete claim set stands exactly as granted (2025-10-28), with no IPR-derived estoppel binding any party and no prior PTAB loss for trinamiX to overcome. For a defendant facing assertion, that is a neutral-to-open position: you are not blocked by any prior petitioner's estoppel and can bring the first ground-floor IPR challenge, but you also cannot point to any existing cancellation as leverage. The absence of IPRs is unsurprising given the patent issued less than a year ago (2025-10-28) and that trinamiX's first U.S. enforcement campaign (against Apple, filed 2026-09-03 in W.D. Tex.) is only days old.

There are no proceedings to profile individually; the remainder of this analysis addresses what that means and what to watch.

Strategic summary

Claim status — CANCELED vs. SUSTAINED vs. UNTESTED. Every claim of US12456334 is UNTESTED in an AIA trial. No petition has been filed, so no claim has been canceled and none has been sustained by a PTAB panel. The patent is Active, granted 2025-10-28, with an anticipated (metadata-assumed) expiration of 2042-02-17. Because the underlying application (US 18/468,352, filed 2023-09-15) is post-AIA and the 9-month post-grant review window closed on 2026-07-28 (nine months after the 2025-10-28 grant), PGR is no longer available — IPR is now the only AIA trial vehicle for this patent.

Estoppel landscape. Because no IPR has ever been filed against US12456334, there is no § 315(e)(2) estoppel running against anyone. Every § 102/§ 103 ground that could have been raised remains fully available to any future petitioner, subject only to that petitioner's own one-year bar under § 315(b) (measured from service of an infringement complaint against it) and to the art being prior to the earliest claimed priority date (2021-02-18, flagged as assumed metadata; the asserted priority chain may matter if trinamiX tries to reach a 2021-02-18 effective filing date — verify against the file history before relying on any 2021-dated reference). No real-party-in-interest or privy web has been created by PTAB filings because none exist.

Pattern signals. There is no repeat-petitioner pattern, no defensive-aggregator presence (e.g., Unified Patents) in the record, and no PTAB appeal history — trinamiX GmbH has not yet defended this patent before the Board because it has never been challenged there. The meaningful signal is on the district-court side: on 2026-09-03, trinamiX (a wholly owned BASF subsidiary) sued Apple Inc. in the U.S. District Court for the Western District of Texas (Midland Division) alleging Face ID (iPhone 15/16/17 series, iPad Pro) infringes seven trinamiX facial-authentication patents focused on material/skin detection. Notably, press reports of that complaint do not enumerate the seven asserted patents, so it is unconfirmed whether US12456334 is among them. This is the case to watch: if US12456334 is asserted against Apple, Apple's § 315(b) IPR window runs one year from service (~2027-09-03), and any Apple IPR would be the first real test of these claims.

Recommended next steps

1. Confirm whether US12456334 is actually in play. Pull the docket of the Midland case (W.D. Tex., filed 2026-09-03; search PACER/Public Access for "trinamiX GmbH v. Apple Inc.") and read the complaint's exhibit listing the seven asserted patents. If US12456334 is not asserted, this patent is presently dormant — a low-immediate-risk profile, but one that can change as trinamiX expands enforcement. The patent's own file wrapper (USPTO Patent Center for US 18/468,352) is the authoritative source for the claim set and priority chain since the claims text was not reproduced in the source material for this analysis.

2. If you are a defendant served with a complaint citing US12456334: your IPR petition deadline is one year from service (35 U.S.C. § 315(b)) — calendar it immediately. PGR is already time-barred (window closed 2026-07-28), so IPR is the only PTAB route. There is no FWD or prior institution decision to cite because none exists — do not let anyone suggest otherwise. A first-filed IPR here carries no § 315(e)(2) baggage: you can run every credible § 102/§ 103 ground you have, including combinations a later petitioner could not reuse after your estoppel attaches.

3. Watch for an Apple IPR. Apple is the natural first petitioner if this patent (or its family members US12288421B2 / US12530925B2 / US12530925B2-continuations) is asserted in the Midland case. Expect any petition between late 2026 and the § 315(b) deadline; institution decisions typically issue ~3 months after filing and a Final Written Decision within 12 months of institution. If Apple petitions and is joined or later settles, its estoppel binds Apple and its privies only — not you.

4. Verify priority before building an invalidity case. The 2021-02-18 priority date is Google Patents metadata (flagged there as an assumption). Confirm via the USPTO file history whether the claims can reach that date or are limited to the 2023-09-15 filing date; a later effective date widens your prior-art pool considerably, particularly for NIR skin-detection/beam-profile references from 2021–2023.

5. No PTAB proceeding exists today — treat that absence as a timing signal, not a merits signal. This patent granted 2025-10-28 and the first enforcement suit landed 2026-09-03; well-asserted patents in active litigation typically attract IPR petitions within the first year of the case. If no petition appears within ~6–9 months of service in the Apple case, that may indicate the accused products' relevant features are hard to map to these claims (weak infringement theories) rather than that the claims are strong — either way, a full prior-art search and claim chart should precede any settlement negotiation.

I was unable to complete full verification of the USPTO Assignment Center records within this session (search quota exhausted before I could pull reel/frame data), so everything below is flagged accordingly. I have not fabricated any assignment entries.

Inventors

All six named inventors are listed on the granted patent and on the family publications as employed by / assigned to trinamiX GmbH, Ludwigshafen am Rhein, Germany:

• Peter Schillen — trinamiX GmbH (listed address: Ludwigshafen am Rhein)
• Benjamin Guthier — trinamiX GmbH (Ludwigshafen am Rhein)
• Friedrich Schick — trinamiX GmbH (Ludwigshafen am Rhein)
• Manuel Guenther — trinamiX GmbH (recorded address in family publication: Zurich, Switzerland)
• Lars Diesselberg — trinamiX GmbH (recorded address: Karlsruhe, Germany)
• Christian Lennartz — trinamiX GmbH (Ludwigshafen am Rhein)

Unusual patterns: none observed. All six inventors appear repeatedly on later trinamiX filings in this same family (e.g., US12288421B2, US12530925B2, continuations filed 2024–2026 per Google Patents legal events), which is inconsistent with a mass-departure / fire-sale pattern. The inventor addresses in Karlsruhe and Zurich are consistent with trinamiX/BASF research locations, not shell-entity addresses.

Original assignee

• Entity named on the issued patent: trinamiX GmbH (Google Patents "Current Assignee"; also the applicant on every family publication, e.g., US20240005703A1, US12530925B2).
• Line of business: trinamiX GmbH is a wholly owned subsidiary of BASF SE, spun out of BASF research around 2015. It develops and commercializes mobile near-infrared (NIR) spectroscopy and biometric sensing technology — including mobile NIR spectroscopy solutions for consumer electronics, NIR sensor modules (e.g., marketed under the "Hertzstueck" brand referenced in the specification), and face-authentication / presentation-attack-detection solutions for smartphone OEMs. It is an operating technology company that both ships/licenses hardware-software solutions, not a paper patent holder.
• Product embodiment: The specification describes the invention as executed on a mobile device (smartphone) camera/illumination stack for face unlock; trinamiX markets face-authentication sensing solutions to device OEMs, so the claims are directed to a product line trinamiX actually sells/licenses.
• Current status: Operating (BASF subsidiary, active as of the fetched record and current date).

Assignment timeline

I could not retrieve a reel/frame-level assignment record for US12456334B2 in this session. USPTO Assignment Center (assignmentcenter.uspto.gov / assignment.uspto.gov) was not directly queryable through the tools available, and web searches surfaced no recorded assignment entries (no reel/frame numbers, no assignor/assignee conveyance records) naming this patent. What the accessible records do show:

• Google Patents legal-events metadata lists only prosecution-family events for this patent — continuations/divisions (US18/743,640 → US12288421B2, filed 2024-06-14; US19/088,110 → US12530925B2, filed 2025-03-24; US19/420,836 → US20260105781A1, filed 2025-12-16) and grant (2025-10-28). These are not ownership conveyances.
• The named assignee on the granted patent and all family publications is unchanged: trinamiX GmbH. No post-issuance transfer to any LLC, holding company, or aggregator is reflected anywhere I could search.
• A standard pre-issuance inventor→assignee assignment very likely exists (it is a practical prerequisite for a German corporate applicant to be named), but I will not invent its reel/frame or date.

Plain statement: No recorded assignments were surfaced. The best available evidence is that trinamiX GmbH (BASF SE subsidiary) still owns US12456334B2 — this is itself a finding, since it means no NPE transfer has occurred on the public record.

Caveat: I cannot affirmatively state the USPTO Assignment Center has zero records because I could not complete a direct database query. Verify at https://assignmentcenter.uspto.gov/ (search by patent number 12456334) before relying on this conclusion.

Timeline diagram

timeline
    title Ownership of US 12456334
    2021 : EP priority filed by trinamiX
    2022 : PCT filed
    2023 : US continuation filed
    2025 : Granted to trinamiX GmbH
    2026 : Still held by trinamiX GmbH

NPE / troll-pattern signals

1. Shell-entity transfer — not present. No recorded transfer of any kind, and certainly none to an "IP / Licensing / Holdings" LLC. The patent sits with an operating BASF subsidiary.
2. Known asserter in the chain — not present. No Acacia, Marathon, IV, Wi-LAN, Conversant, RPX, or Unified Patents-listed entity appears anywhere in the ownership or legal-events record. trinamiX GmbH does not appear on public NPE/asserter directories as a licensing-only entity.
3. Repeat correspondent across the chain — not present / unclear. The prosecuting counsel of record for the family is Armstrong Teasdale LLP (listed as the agent/attorney for family member US12530925B2; the "Patent Docket Department" line on US20240037201A1 is consistent with an Armstrong Teasdale docket department entry). A single prosecuting firm for a German corporate applicant is entirely normal and is not an NPE tell; there is no multi-link assignment chain in which a correspondent recurs.
4. Cascading transfers — not present. There are no consecutive assignments through chained LLCs; the only chain is the continuation family, which stays inside trinamiX.
5. Pre-litigation transfer — unclear. Search results around 2026-09-04 referenced reports that trinamiX/BASF sued Apple over Face ID in the Western District of Texas, but the asserted patent numbers were not enumerated, and no CAFC docket referencing US12456334 was found. Even if that suit includes this patent, an operating-company suit against an implementer is not an NPE "pre-litigation transfer" signal — no transfer occurred.
6. Bankruptcy fire-sale — not present. BASF SE (parent) is investment-grade and operating; trinamiX shows no bankruptcy proceeding.
7. Privateering — not present. There is no operating-company-to-NPE handoff; trinamiX would be asserting (if at all) in its own name as a BASF subsidiary.
8. Defensive aggregator (anti-NPE) — not present. The chain does not terminate at RPX/AST/LOT/Unified/OIN; it remains with the original operating assignee.

Verdict

Insufficient data.

The accessible record shows no assignments at all — the patent remains with the original operating assignee, trinamiX GmbH, a wholly owned BASF SE subsidiary that develops and sells/licenses NIR spectroscopy and face-authentication sensing products. No shell-entity transfer, no known-asserter appearance, no cascading LLC chain, and no bankruptcy context exists on the public record, so none of the NPE signals can be marked present. The only reason this is not a confident "operating-company assertion" or "defensive / non-asserting" call is that I could not complete a direct USPTO Assignment Center reel/frame query and could not confirm whether the reported trinamiX/BASF v. Apple W.D. Tex. litigation actually asserts this specific patent. Verify the assignment record at https://assignmentcenter.uspto.gov/ (patent 12456334) to confirm the absence of recorded conveyances.

I'll research U.S. Patent No. 12,456,334 (US12456334B2) and its cited prior art references. Let me run several targeted searches.

I found the family details. Now let me look specifically at the citations (References Cited) listed on the granted patent's front page/record.

The direct citation listing isn't surfacing. Let me search the USPTO Patent Center/assignment records and Google Patents citation pages, plus the known references mentioned in the specification.

I've reached the limit of searches available for this session, so I'll compile the analysis from what was retrievable and flag clearly what could not be verified.

Important caveats up front

1. The "References Cited" section of the granted patent (US12456334B2) is not in the patent text provided to me, and my web searches did not return the USPTO/Google Patents front-page citation list for this specific grant. Google Patents typically renders this as a "Patent Citations" / "Non-Patent Citations" block; that block was absent from the fetched text, and targeted searches did not surface it. The analysis below therefore identifies (a) references affirmatively discussed in the specification of US12456334B2 (which are strong candidates for the cited art) and (b) confirmed family/publication records. This is a reconstruction, not the verified examiner's citation list. If you need the exact 37 CFR 1.56/1.98 list, pull the "References Cited" page from USPTO Patent Center for application 18/468,352.

2. The verbatim claims of US12456334B2 were not reproduced in the source text. The independent method claim reconstructed from the specification (and confirmed verbatim in the sibling publications US20240037201A1 / US12530925B2) requires four conjunctive steps: (a) 2D face detection via pre-defined geometrical features; (b) skin detection by projecting an illumination pattern having a plurality of illumination features, capturing a second image whose reflection features each have a beam profile, deriving first beam-profile information for at least one reflection feature inside the facial image region, deriving a material property from that beam-profile information, and characterizing the face as skin if the material property is characteristic of skin; (c) 3D detection by deriving second beam-profile information for at least four reflection features inside the same facial image region, determining a depth level, and characterizing the face as a 3D object if depth deviates from a plane-object level; and (d) authenticating only if the face is characterized as both skin and 3D. § 102 anticipation analysis below is against that reconstructed claim and assumes the issued claims match the family's published claims — which should be verified.

3. Literal-identifier rule: The specification cites "US 2019/213309 A1." Taken literally that string is not a valid publication number; the description it accompanies (ranging sensor = time-of-flight sensor + reflectance sensor; processor activates facial recognition in response to distance and reflectivity) matches US20190213309A1 exactly (STMicroelectronics, Facial authentication systems and methods utilizing time of flight sensing), confirmed by search. I treat them as the same reference but note the discrepancy.

Prior art references relevant to US12456334B2

1. US20190213309A1 — STMicroelectronics, Inc.

• Full citation: US 2019/0213309 A1, Facial authentication systems and methods utilizing time of flight sensing, inventors (per record) assigned to STMicroelectronics, Inc.; application filed 2018-01-05; published 2019-07-11. (Cited in the specification as "US 2019/213309 A1.")
• Description: Face-authentication electronic device with a ranging sensor (ToF), a reflectance sensor, an RGB camera, an ambient-light sensor, and a beam generator. The ToF sensor measures distance and, with the reflectance sensor, a skin-reflectance value of the object; a processor activates a facial-recognition process in response to the distance and reflectivity (e.g., determines that the reflectance value is that of human skin and the user is within a recognition range), and can create a depth map of a face. Two power modes (single-point presence detection; multi-point face/depth-map acquisition).
• § 102 analysis: This is the reference the applicants themselves distinguished. It discloses step (a)-type face detection and a reflectance-value skin check plus ToF depth mapping, but it does not disclose: projecting an illumination pattern of a plurality of illumination features; analyzing the beam profile of individual reflection features to derive a material property (skin vs. non-skin materials like masks/prints); deriving depth from beam-profile information of at least four reflection features (its depth comes from ToF, not beam-profile/DPR analysis); or an authentication step gated on the conjunction of "skin" and "3D object." A single-reference § 102 anticipation of the full method claim therefore fails (missing limitations b–c's beam-profile/material analysis and the conjunctive authentication gate). It is the strongest background art for a § 103 obviousness combination (e.g., STM's skin-reflectance + ToF concept combined with trinamiX's beam-profile analysis), and it could anticipate any dependent claim that happened to be drawn narrowly to "reflectance-based skin detection + ToF depth" — but no such issued claim is evident.

2. WO 2018/091649 A1, WO 2018/091638 A1, WO 2018/091640 A1 — trinamiX GmbH (Depth-from-Photon-Ratio family)

• Full citation (representative): WO 2018/091649 A1, Detector for determining a position of at least one object (and sibling publications WO 2018/091638 A1, WO 2018/091640 A1), published 2018-05-24; priority 2016-11-17. The US12456334B2 specification incorporates these by reference as the "depth-from-photon-ratio (DPR) technique."
• Description: Optical detectors/methods determining a longitudinal coordinate (depth) of an object by projecting a light beam and analyzing the beam profile of the reflection feature on a sensor — specifically by taking first and second areas of the beam profile (center vs. edge information, e.g., trapezoid beam profiles) and forming a quotient Q, then mapping Q to distance via a predetermined relationship. Discloses the beam-profile/trapezoid mathematics, center/edge integrals, and spot-shape-based ranging that step (c) of US12456334B2 uses.
• § 102 analysis: As publications from May 2018 they are prior art under AIA § 102(a)(1)/(2) against the 2021-02-18 priority date. They disclose the sub-technique embedded in step (c) — determining depth from second beam-profile information of reflection features — and could anticipate a claim drawn solely to that DPR ranging method. But they are not directed to face authentication: they lack the face-detection step (2D geometrical features), the skin/material-classification step, the "at least four reflection features inside the facial image region" facial-region limitation, and the conjunctive authentication step. Not anticipatory of the full method claim.

3. WO 2020/187719 A1 — trinamiX GmbH (material/beam-profile classification)

• Full citation: WO 2020/187719 A1, published 2020-09-24 (PCT claiming priority from trinamiX EP applications of early 2020). The US12456334B2 specification cites it (with EP 20159984.2 filed 2020-02-28 and EP 20 154 961.5 filed 2020-01-31, incorporated by reference) for "beam profile analysis… to classify materials," including distinguishing human skin from other materials via back-scattering/sub-surface-scattering properties.
• Description: Methods of classifying a material's surface by projecting light (e.g., a laser spot) and analyzing the reflection/back-scattering beam profile — surface (Lambertian-like) reflection vs. diffuse sub-surface scattering — with feature-based filters (luminance, spot-shape, smoothness/speckle-contrast, grey-level-occurrence, Law's energy, threshold-area) optionally with machine learning/CNN classifiers. This is the technology underlying step (b) of US12456334B2 (determining a material property from first beam-profile information to characterize skin).
• § 102 analysis: Published ~Sept 2020, before the 2021-02-18 priority date, so it is § 102 prior art. It plausibly anticipates the skin/material-classification sub-combination of step (b), but it does not disclose face detection, the facial-image-region mapping, 3D detection with ≥4 reflection features, or the authentication gate. Not anticipatory of the full method claim. Note: I could not confirm from this session whether WO 2020/187719 is in the examiner's formal citation list versus merely incorporated-by-reference text; treat this as a specification-identified reference.

4. EP 20 154 961.5 (filed 2020-01-31) and EP 20159984.2 (filed 2020-02-28) — trinamiX GmbH (as incorporated by reference)

• Full citation: European patent applications EP 20 154 961.5 (filed 2020-01-31) and EP 20159984.2 (filed 2020-02-28), both incorporated by reference in the US12456334B2 specification for skin/beam-profile material classification (the latter also cited in family text as the basis of the PAD skin-classification approach).
• Description: European filings underlying the material-classification-by-beam-profile technology later reflected in WO 2020/187719.
• § 102 analysis: European applications publish ~18 months after filing (i.e., ~mid-2021), which is after the 2021-02-18 priority date of US12456334B2; unless an earlier-published counterpart (e.g., the PCT application that became WO 2020/187719) qualifies, these EP filings themselves are likely not prior art under § 102(a)(1)/(2) as of the effective filing date. They are more relevant as priority/incorporation disclosures than as anticipatory references.

5. EP 20 170 905.2 (filed 2020-04-22) — trinamiX GmbH (hexagonal laser-spot grid)

• Full citation: EP application 20 170 905.2, filed 2020-04-22, incorporated by reference in US12456334B2 for projecting "a periodic grid of laser spots, e.g., a hexagonal grid."
• Description: Illumination-pattern generation (periodic/hexagonal spot grid) for beam-profile-based ranging/material analysis.
• § 102 analysis: Same timing issue as item 4 — likely unpublished as of 2021-02-18 and thus not § 102 prior art in its own right; relevant to the illumination-pattern limitation of step (b) only through its incorporated disclosure.

6. Non-patent literature identified in the specification

• Masi, Iacopo, et al., "Deep face recognition: A survey," 2018 31st SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI), IEEE, 2018 — incorporated for the face-detection/recognition techniques of step (a). Published 2018 ⇒ § 102(a)(1) prior art as to any claim limitation directed to known 2D face-detection algorithms (geometrical feature identification), but not anticipatory of the combined method.
• Jürgen Eichler & Theo Seiler, "Lasertechnik in der Medizin: Grundlagen, Systeme, Anwendungen," ch. "Wirkung von Laserstrahlung auf Gewebe," Springer Verlag, 1991, pp. 171–266, ISBN 0939-0979 — cited for the human-skin reflection profile (surface reflection + penetration-depth-dependent diffuse back-scattering vs. wavelength). 1991 publication ⇒ prior art; foundational science for the skin/material distinction in step (b), not anticipatory standing alone.

Bottom line on § 102 anticipation

• No single reference located in this session appears to anticipate the full reconstructed independent method claim (steps a–d), because each candidate lacks at least one conjunctive element — most commonly the combination of (i) beam-profile-derived material classification (skin vs. non-skin) and (ii) beam-profile-derived 3D/depth detection of ≥4 facial reflection features and (iii) authentication gated on the skin-AND-3D conjunction. US20190213309A1 lacks all of (i)–(iii) in the claimed form; the trinamiX DPR publications (WO 2018/091649, /091638, /091640) lack (i) and the face-authentication context; WO 2020/187719 lacks the face/3D/authentication context.
• The most credible § 102 threats are to sub-combination/dependent claims: WO 2020/187719 (and its EP priority filings, if an earlier publication exists) against skin/material-classification claims; the WO 2018 DPR trio against depth-from-beam-profile claims; US20190213309A1 against any claim reciting reflectance-gated face recognition or ToF depth.
• The strongest attack on the independent claim is likely a § 103 obviousness combination (e.g., US20190213309A1 in view of WO 2020/187719 and/or WO 2018/091649), not a single-reference § 102 anticipation — but that depends on the exact issued claim scope, which I could not verify from the provided record.
• Unverified items to confirm against USPTO Patent Center (application 18/468,352): the examiner's complete "References Cited" list (which may include additional art not discussed in the specification, e.g., other face-PAD patents such as Samsung/Apple Face ID art), the issued claim set, and any Information Disclosure Statement filed by TrinamiX listing the EP/WO family documents.

If you can provide the "References Cited" page from the issued patent or Patent Center, I can complete the element-by-element § 102 chart for each listed reference.

§ 103 Obviousness Analysis — U.S. Patent 12456334 ("Optical skin detection for face unlock")

0. Scope caveat (read first)

The granted claims of US12456334B2 were not reproduced in the source text provided to me, and the earlier sections of this analysis flagged that the claim set had to be reconstructed from the specification. The specification states the invention solves its problem "with the features of the independent patent claims," and the family members located during research (e.g., US12288421B2 at patents.justia.com/patent/12288421; the publication US20240005703A1 at patents.justia.com/patent/20240005703) mirror the same representative method claim:

• (a) face detection step — determine a 2D first image of a scene; detect the face by identifying predefined/pre-determined geometrical features characteristic of faces, using a processing unit;
• (b) skin detection step — project an illumination pattern of multiple illumination features; determine a second image of reflection features each having a beam profile; for at least one reflection feature inside the image region corresponding to the detected face, analyze the beam profile to obtain first beam-profile information; determine a material property from it; characterize the face as skin if the material property corresponds to a skin property;
• (c) 3D detection step — for at least four reflection features in the same facial image region, analyze beam profiles to obtain second beam-profile information; determine at least one depth level; characterize the face as a 3D object if the depth level deviates from the depth level of plane objects;
• (d) authentication step — authenticate only if the face is characterized as skin and as a 3D object.

A sibling family member also claims the corresponding mobile device (illumination unit + camera + processing unit executing steps ii/iii/a) and uses. The analysis below therefore proceeds against this representative claim scope, with dependent-feature notes in § 7. If the granted claims were amended to narrower, unanticipated subject matter, the analysis must be re-run against the actual claim text from the USPTO file wrapper.

1. Graham framework and the person of ordinary skill

Under Graham v. John Deere, 383 U.S. 1 (1966), obviousness turns on (1) the scope and content of the prior art, (2) differences between the prior art and the claims, (3) the level of ordinary skill, and (4) secondary considerations.

Level of ordinary skill. A POSITA would be a team: an image-processing/computer-vision engineer experienced in face detection and biometric authentication, and an optical-sensor engineer experienced in active NIR illumination, structured-light/depth sensing, and reflection/scattering analysis for mobile devices (smartphones). The relevant art — face unlock with presentation-attack detection (PAD) — was crowded by the priority date (Feb. 18, 2021): Apple's Face ID shipped in 2017; Qualcomm, Intel, STMicroelectronics, Goodix and others had published PAD designs; and the applicant's own trinamiX detector families (2018–2020) had published beam-profile-based depth and material classification.

2. Prior-art inventory (from the prosecution/citation record and the specification itself)

The Google Patents page for the pre-grant publication of this patent (US20240005703A1, the publication of application 18/468,352 that issued as US12456334B2) lists examiner/author citations — "Citations (4)" and "Family Cites Families (8)" — found at https://patents.google.com/patent/US20240005703A1/en#21, supplemented by the specification's own designated background art:

Note on dates: All of the above were published or filed (and for § 102(a)(2) purposes, sufficiently early) before the Feb. 18, 2021 priority date. trinamiX's own WO2018/091649 (2018) and WO2020/187719 (Sept. 2020) publications are prior art against this patent even though they share an applicant; same-owner publications are not disqualified under § 103.

3. The differences between the claims and the prior art (framing the real question)

Individually, no single reference discloses the full four-step pipeline. The genuinely contested features are concentrated in step (b): classifying skin by deriving a material property from the beam profile of a reflected illumination feature located within the detected facial region, and the AND-gating of step (d) on both skin and 3D-ness. Everything else — 2D face detection (a), multi-point depth/3D detection by analyzing ≥4 projected spots inside the face (c), and conditional authentication (d) — was well developed in the art by 2021.

Critically, the patent itself concedes the enabling technology was pre-existing: it incorporates the trinamiX DPR applications (WO2018/091649A1, /638, /640) by reference for depth and the trinamiX material-identification application (WO2020/187719, EP 20159984.2, EP 20 154 961.5) by reference for "classifying materials … [including] human skin," and it states that "beam profile analysis simultaneously provides depth information by analyzing the same camera frame." The only increment over trinamiX's own earlier published detectors is (i) restricting the analysis to reflection features falling in a face-detected image region and (ii) gating face authentication on the two conditions. Both increments were already motivated by the face-auth PAD art (US 2019/213309 A1, Qualcomm, Intel).

4. Combination 1 (primary): US 2019/213309 A1 (or US10922395B2) + trinamiX WO2020/187719 + trinamiX WO2018/091649 family (+ routine face detection)

Map. US 2019/213309 A1 supplies the context: a face-authentication system in which a processor activates facial recognition only after evaluating a ranging signal (distance) and a reflectance signal from the user — i.e., it already discloses the concept of gating/authenticating on both a geometric/distance check and a reflection/material-ish check, and its own stated limitation (per the patent's background) is that it "detects simple presentation attacks with photographs and videos … [but] an approach to reliably detect presentation attacks with 3D face masks is still missing." Substituting/addressing that exact limitation with (1) WO2020/187719's beam-profile material classification (which classifies skin vs. non-skin and volume-scattering materials) and (2) the DPR depth measurement of WO2018/091649 (which derives per-spot distance from the same beam-profile analysis) yields the claimed method: analyze the reflection feature's beam profile → material property (b); analyze ≥4 features' beam profiles → depth level/3D topology (c); authenticate only when both pass (d). Face detection (a) is supplied by any standard detector (Masi survey; Qualcomm's landmark detector; Intel's face detection).

Why a POSITA would combine.

• Same field, same problem: All three references are in biometric/optical sensing; US 2019/213309 A1 explicitly frames the unsolved problem (3D-mask spoofs) that WO2020/187719's material channel and the DPR depth channel solve.
• Known interchangeability and common origin: DPR and material classification are disclosed by the same applicant as two uses of one measurement principle (beam-profile analysis of reflected spots). A POSITA combining them does no more than use two known functions of the same underlying signal.
• Express technological motivation: The specification itself describes DPR and material classification as deriving from the same camera frame — the motivation to fuse them (speed, one sensor, one frame) is inherent in the prior art's own teaching that both are beam-profile techniques.
• Predictable result: Classifying skin by scattering/translucency and measuring 3D shape by spot geometry were each known; AND-gating an authenticator on two independent anti-spoof channels was already the design pattern of US 2019/213309 A1 (distance + reflectivity) and of Intel/Qualcomm (multi-test liveness). This is a classic KSR combination of known elements each performing its known function.

5. Combination 2 (strong): Qualcomm US20210049391A1 + Intel US20160335483A1 (+ standard face detection)

Map. Qualcomm discloses a complete phone-unlock/face-authentication pipeline: capture image of a face → detect facial landmarks (element (a)) → obtain a depth image of the face → map landmarks to depth → determine liveness from depth structure (element (c) in substance: 3D structure vs. flat spoof) → gate authentication (element (d) in part). Qualcomm's own disclosure identifies its target as photo/tablet/phone (2D) spoofs — precisely the class the patent admits "can be detected" by the prior art. Intel '835483 discloses the missing material/skin channel: active illumination (including IR), analysis of how an emitted pattern is reflected — including reflectivity thresholds distinguishing live body parts (IR reflects off skin) from facsimiles (screens/photos), and analysis of pattern deformation on contoured (3D) surfaces vs. flat spoof media — and denial of authentication on a spoof indication. Intel also teaches applying a set of liveness tests ("each test in the set … may include an assay that differs between a live face and a face simulated on a pictorial representation"), i.e., stacking independent liveness channels, exactly the architecture of claim steps (b)+(c)+(d).

Why a POSITA would combine. A designer of Qualcomm-style depth-liveness face unlock who reads Intel learns (i) that depth alone is defeated by curved/3D presentation media and (ii) that a material/reflectivity assay adds an orthogonal channel. The combination — require both the Qualcomm depth/liveness condition and the Intel skin/reflection condition before authenticating — directly produces the claimed method. Both references are in the identical art unit (facial liveness/PAD for mobile authentication), both target spoofing of face unlock, both are expressly combinable because Intel teaches multi-assay liveness stacking and Qualcomm's architecture has a natural insertion point for a per-landmark material check (Qualcomm already maps landmarks onto a second modality image, i.e., it teaches fusing a landmark/face-region location with an auxiliary sensor image — the same "image region of the second image corresponding to the … first image" concept of the claims).

6. Combination 3: Apple US10990805B2 + Goodix CN111164610B + trinamiX WO2020/187719 (device claim)

For the mobile device claim (illumination unit + camera + processing unit), Apple's hybrid flood/dot-projector face-authentication hardware on a phone supplies the exact physical architecture: a camera and an illumination unit projecting multiple illumination features for 2D + depth face recognition with spoof rejection. Goodix supplies anti-spoof real-face sensing in the same hardware context. Adding trinamiX WO2020/187719's beam-profile material classifier to the already-present dot-projector camera frame gives the skin channel without any new hardware. Motivation: defense-in-depth against masks, at near-zero marginal cost and latency, using a frame the device already captures — the same rationale Apple/Goodix/Qualcomm each articulate for adding anti-spoof channels, and the same rationale the patent itself gives (material + 3D "from the same frame with low computational effort").

7. Dependent-feature-level notes

• Machine-learning/parametrized skin classification model (spec Embodiment 4; dependent claims): WO2020/187719 already contemplates classification models, and the patent itself says "convolutional neural networks may be utilized to classify skin by using the reflection images as an input" — a routine implementation choice. Training classifiers on skin vs. mask materials was standard PAD practice (Intel, Qualcomm, and the Goodix art all use learned classifiers).
• Periodic grid of laser spots / NIR ~940 nm: dot-projector grids are disclosed by Apple US10990805B2 and the trinamiX EP 20 170 905.2 family incorporated in the specification; 940 nm is a conventional VCSEL wavelength whose melanin-insensitivity is textbook optics (the specification itself explains the physics), making wavelength selection an obvious design choice for skin-tone agnosticism.
• Secure-zone execution and <500 ms timing: performance constraints of this kind are standard design goals for mobile biometrics and would not independently support patentability.

8. Countervailing considerations (the honest weaknesses of the obviousness case)

A well-advised patentee would press the following:

1. No single reference teaches "skin classification from beam-profile (spot-shape/translucency) analysis of reflected laser spots" in a face-authentication context. Intel's reflectivity assays are threshold-based and pattern-homogeneity-based, not beam-profile/material-classification based as the claims define it. WO2020/187719 teaches material identification generically (e.g., for consumer/material sensing), and a challenger must supply an articulated reason to transplant it into face unlock; while Combination 1 provides that reason via US 2019/213309 A1's acknowledged mask problem, the three-reference combination is more complex than the typical obviousness case and may invite an "impermissible hindsight" argument unless the motivation is anchored in the references themselves.
2. Non-obviousness of the fused single-frame pipeline — deriving both the material property and the depth from the same reflection features in the same second image, restricted to the face region — is not squarely disclosed in any one reference and produces a real (if incremental) efficiency/speed benefit. If the granted claims capture that fusion with specificity, the obviousness case weakens.
3. Region-specificity ("at least one reflection feature located inside an image region … corresponding to the identified geometrical feature," and the ≥4-feature face-region requirement for step (c)) is a spatial-selection detail that, while it reads as an obvious implementation choice (Qualcomm's landmark-to-depth mapping and Intel's face-marker analysis both localize analysis to the face), gives the examiner/patentee room to argue the prior art analyzed whole scenes rather than facial regions.
4. Secondary considerations: if the trinamiX/BASF v. Apple W.D. Tex. litigation (reported Sept. 2026, per the earlier litigation summary) proceeds and commercial-success evidence with a nexus to the claimed features is developed, that evidence would be weighed — though note that the asserted commercial success of Face ID would cut against a nexus to this patent's features.

On balance, however, the dominant prior-art narrative is unfavorable to patentability of the representative claim at the level of generality reconstructed here: the specification itself concedes that (i) 2D spoofs were detectable by US 2019/213309 A1-type systems, (ii) the missing piece — 3D-mask detection — is supplied by the combination of trinamiX's own pre-priority DPR (WO2018/091649) and material-classification (WO2020/187719) publications, and (iii) the "same frame" efficiency is a natural consequence of both being beam-profile techniques. Adding face-region localization (Qualcomm/Intel) and AND-gating (US 2019/213309 A1, Intel multi-assay liveness) to that known core is the kind of "combination of familiar elements according to known methods" that KSR International Co. v. Teleflex Inc., 550 U.S. 398 (2007), treats as obvious when it "does no more than yield predictable results."

9. Bottom line

Most likely to succeed (highest obviousness risk): A § 103 challenge combining US 2019/213309 A1 (face-authentication gated on distance + reflectance; acknowledged 2D-spoof detection; identified mask-spoof gap) with WO2020/187719 (beam-profile material/skin classification) and the WO2018/091649 DPR family (beam-profile depth from ≥1 spot), plus a conventional face detector (Masi/Qualcomm/Intel) — the trinamiX DPR and material references being same-applicant but valid prior art published in 2018 and Sept. 2020, respectively, before the Feb. 18, 2021 priority date. A close second is Qualcomm US20210049391A1 + Intel US20160335483A1, which together disclose the full architecture (2D face detection, landmark-localized depth-liveness analysis of the face, active-illumination reflection/skin analysis, multi-assay liveness gating of authentication) and therefore render the claimed combination little more than the union of two references in the identical field, addressing the identical spoofing problem, with the identical AND-gate logic.

Confidence is qualified by the missing granted-claim text; this analysis should be validated against the actual claims (USPTO Patent Center for US12456334B2) before it is relied upon in any pleading or opinion.