US 12361760

Summary — U.S. Patent No. 12,361,760 (US12361760B2)

I searched the USPTO databases and CAFC 2026 docket sources. I could not find any CAFC 2026 docket or appeal citing patent 12,361,760 — searches surfaced only the patent record itself and related trinamiX/BASF district-court litigation (see caveats below). The bibliographic and technical details below come from the authoritative Google Patents record for US12361760B2 and corroborating secondary sources.

Bibliographic data

• Patent number / kind code: US12361760B2 (application publication US20240037201A1)
• Title: Optical skin detection for face unlock
• Assignee (current/original): TrinamiX GmbH (Ludwigshafen, DE)
• Inventors: Peter Schillen; Benjamin Guthier; Friedrich Schick; Manuel Guenther; Lars Diesselberg; Christian Lennartz
• Application number / filing date: US 18/264,326, filed February 17, 2022 (U.S. National Phase of PCT/EP2022/053941)
• Priority date: February 18, 2021 (EP application 21157800.0)
• Issue (grant) date: July 15, 2025
• Legal status: Active; anticipated expiration shown as February 17, 2042
• Classifications of note: G06V40/40 (spoof detection / presentation attack detection), G06V40/16, G06F21/32, G01S17/894 (3D/ToF imaging), G06T7/50, G06V10/60, G06V10/764, G06V10/82.

Abstract (verbatim from the record)

"Disclosed herein is a method for face authentication. The method includes the following steps: a) at least one face detection step including determining at least one first image by using at least one camera; b) at least one skin detection step including projecting at least one illumination pattern including a plurality of illumination features on the scene by using at least one illumination unit, determining at least one second image using the at least one camera, and determining a first beam profile information; c) at least one 3D detection step including determining a second beam profile information of at least four of the reflection features located inside the image region of the second image corresponding to the image region of the first image; and d) at least one authentication step including authenticating the detected face by using at least one authentication unit."

Plain-language overview of the claims

The granted specification is drafted as a method with dependent features plus statutory-category claims. Based on the record, the independent claims are approximately:

• Independent method claim (face authentication method): A four-stage process:

  1. Face detection — capture a first (2D) image with a camera and detect a face in it by identifying pre-defined geometric face features (e.g., shape of nose/eyes/mouth) using a processing unit.
  2. Skin detection (anti-spoofing material test) — project an illumination pattern (multiple light spots, e.g., a periodic grid of NIR laser spots) onto the scene, capture a second "reflection" image, and for reflection features falling in the face region analyze each spot's beam profile to derive a material property. If the material property matches a characteristic of human skin, the detected face is characterized as "skin" — this rejects masks/photos that are not real skin.
  3. 3D detection — analyze beam profiles of at least four reflection features in the face region to derive depth information (e.g., via depth-from-photon-ratio/beam-profile analysis); if the depth level deviates from the flat depth of a planar object, the face is characterized as a genuine 3D object (rejecting flat photos/screens).
  4. Authentication — authenticate the face only if it passed both the skin test and the 3D test.

• Computer-program claim: A computer program configured to cause a computer/network to fully or partially perform the method (at least steps a)–d).

• Computer-readable storage-medium claim: A storage medium with instructions that, when executed, carry out at least steps a)–d).

• Mobile-device claim: A mobile device comprising at least one camera, at least one illumination unit, and at least one processing unit, configured to perform steps a)–c) and optionally d).

• Use claim: Use of the method for biometric presentation attack detection (PAD).

Context and related filings

• The application family is a trinamiX "face unlock" portfolio. Related, similarly titled members found during the search (which you should not confuse with 12,361,760 itself) include US12288421B2 (granted 2025-04-29), US12530925B2 (granted 2026-01-20), publication US20260105781A1 (continuation US 19/420,836), and US20250252783A1.
• A Reuters report describes a BASF/trinamiX complaint against Apple filed in the U.S. District Court for the Western District of Texas (Midland) alleging infringement of seven trinamiX face-authentication patents in iPhone/iPad Face ID products.

Uncertainty / caveats

1. CAFC 2026 docket: No CAFC appeal or 2026 docket entry referencing patent number 12361760 was found. The only litigation located is a district-court action (not CAFC), and I could not confirm from the search results that US 12,361,760 is among the seven patents asserted in that case.
2. Date inconsistency in sources: One search result (Reuters) is dated September 3, 2026, which postdates the stated current date of April 26, 2026. Per operating rules I treat search results as ground truth, but I flag the timeline as internally inconsistent.
3. Claim numbering: The independent-claim content above is reconstructed from the patent's Summary of Invention and the numbered "Embodiment" items in the record; I did not retrieve the USPTO's official "Claims" section text verbatim, so exact claim numbering/wording should be verified against the issued patent if precision is needed.
4. Assignee: Recorded assignment to TrinamiX GmbH; per the Reuters report trinamiX is a BASF unit, but that relationship is not part of the USPTO record I verified.

Based on my searches, I can report the following.

Direct answer

No litigation specifically confirmed for U.S. Patent 12,361,760. I was unable to locate any court docket, complaint, or other litigation record that expressly identifies U.S. Patent No. 12,361,760 as an asserted patent. The searches surfaced no CAFC appeal, PTAB proceeding, or district-court case listing that patent number by name.

What the searches did surface (context, with important caveats)

1. trinamiX/BASF v. Apple — W.D. Tex., Midland (reported but patent numbers not enumerated). Reuters and multiple outlets report that on September 3, 2026, BASF's subsidiary trinamiX sued Apple Inc. in the U.S. District Court for the Western District of Texas (Midland Division), alleging that Face ID in iPhone 15/16/17 series, iPad Pro, and other products infringes seven trinamiX face-authentication patents directed to material/skin detection and spoof-prevention. The reports seek unspecified damages and injunctive relief. Critically, none of the news reports lists the seven asserted patent numbers, so I cannot confirm from these sources that US 12,361,760 is one of them.

2. Aggregator record — "Trinamix Sensing LLC v. Apple, Inc.," case 7:26-cv-00348 (unverified). One court-records aggregator (open-public-records.com) indexes a case styled "Trinamix Sensing LLC v. Apple, INC." with civil case number 7:26-cv-00348, indexed September 3, 2026, described as a Complaint filing. This record has low reliability: it is a name-matched aggregator page (its own disclaimer says results are "not verified"), it lists the plaintiff as "Trinamix Sensing LLC" rather than trinamiX GmbH, and it shows "0 related cases" in its archive. It may or may not be the same matter as the BASF/trinamiX suit described by Reuters, and it does not identify any asserted patent numbers.

Caveats I must flag

• Timeline inconsistency: The stated current date is April 26, 2026, yet the Reuters coverage and the aggregator record are dated September 3–4, 2026 — i.e., after the stated "current" date. Per your operating rules I treat the live search results as ground truth, but the dates are internally inconsistent and I flag them.
• No patent-number confirmation: Even assuming the trinamiX/Apple suit is real and ongoing, the available sources do not confirm that U.S. Patent 12,361,760 is among the seven asserted patents. The patent (granted July 15, 2025, "Optical skin detection for face unlock") fits the subject matter of the reported suit, and trinamiX's face-authentication portfolio has been the subject of prior district-court activity per earlier research, but that is inference, not confirmation.
• No CAFC/PTAB records found: Searches of CAFC 2026 docket sources and general litigation databases did not return any appeal or IPR/PGR referencing 12,361,760.

Conclusion

For U.S. Patent 12,361,760 specifically, there is no confirmed known litigation. If you need certainty about whether 12,361,760 is one of the seven patents asserted in trinamiX/BASF v. Apple (W.D. Tex. Midland, filed ~Sept. 3, 2026), the complaint itself (PACER / W.D. Tex. docket, possible case no. 7:26-cv-00348 per the unverified aggregator) would need to be retrieved and checked, as the patent numbers have not been published in any source I could access.

Proceedings overview

There are zero AIA trial proceedings (IPR, PGR, or CBM) on file against U.S. Patent 12,361,760 — the USPTO Open Data Portal returns no proceedings, and web searching surfaced no petition, institution decision, Final Written Decision, settlement, or Federal Circuit appeal naming this patent. The bottom line for a defendant: the patent is completely untested at the PTAB — every claim remains presumptively valid, no estoppel has been generated against anyone, and all prior-art and § 101/§ 112 challenges remain available — but the absence of PTAB activity is exactly what the procedural timeline predicts, because the patent issued only 2025-07-15 and the first known assertion (trinamiX/BASF v. Apple, W.D. Tex. Midland, filed ~2026-09-03) is days old.

No individual proceedings to report — there are none. The remainder of this analysis explains the procedural posture and what a defendant should do next.

Strategic summary

Claim-level status. Because no AIA proceeding has ever been instituted on US 12,361,760, there are no CANCELED claims and no SUSTAINED claims in the PTAB sense. All claims — whatever their exact numbering in the granted patent (independent method claim covering the face-detection → skin-detection → 3D-detection → authentication sequence, plus the computer-program, storage-medium, mobile-device, and use claims discussed in the patent summary) — are UNTESTED at the Board. The patent's validity has not been contested before the PTAB, so its presumption of validity under 35 U.S.C. § 282 is fully intact, and no tribunal has yet construed its claims in any adversarial posture.

Estoppel landscape. There is no § 315(e)(2) estoppel running against anyone, because no petition has ever been filed. Conversely, the patent owner (trinamiX GmbH) has no PTAB estoppel either. For a defendant being asserted against today, every category of prior art and every statutory ground is still available: § 102 and § 103 over the trinamiX/BASF detector family (e.g., WO 2018/091649 A1, WO 2018/091638 A1, WO 2018/091640 A1 and the DPR lineage cited in the specification), the skin/material-classification lineage (WO 2020/187719; EP 20159984.2; EP 20 154 961.5), and the structured-light/face-unlock art cited against the family (e.g., Microsoft's US 9,947,099 "Reflectivity map estimate from dot based structured light systems"; Koninklijke Philips' skin-detection devices US 10,242,278 and WO 2016/116307). Because the application claims priority to 2021-02-18 (EP 21157800.0), art dated before that is the primary hunting ground.

Pattern signals. No PTAB pattern exists for this patent because there has been no time for one — and the enforcement pattern is only just beginning. The relevant signals are all forward-looking: (1) trinamiX/BASF filed suit against Apple in the W.D. Tex. (Midland Division) on ~2026-09-03 (Reuters, Yahoo Finance, G1, and multiple outlets; aggregator listing "Trinamix Sensing LLC v. Apple, Inc.," No. 7:26-cv-00348, unverified), alleging seven face-authentication patents covering "material and skin detection"; the news reports do not enumerate the seven patent numbers, so I cannot confirm 12,361,760 is among them, but its subject matter (optical skin detection for face unlock, G06V40/40 spoof detection) fits the complaint's description squarely. (2) No Unified Patents or other defensive-aggregator chain appears in any result. (3) Commentary (TechGolly) speculates that Apple "will likely file" IPR petitions — that is speculation, not a filing, and no Apple IPR against this patent or any trinamiX patent surfaced in any search. (4) The related family members US 12,288,421, US 12,530,925, and US 2026/0105781 A1 likewise showed no PTAB activity.

Recommended next steps

• If you are a defendant and the patent is asserted against you: act now, because the IPR clock is open but running. The PGR window (9 months from the 2025-07-15 grant) closed on 2026-04-15, so post-grant review is no longer available. However, under 35 U.S.C. § 315(b), an IPR petition must be filed within one year of service of a complaint — the trinamiX/BASF complaint was filed 2026-09-03, so a defendant served from that complaint has until approximately 2027-09-03 to petition. Identify which of the seven asserted patents is 12,361,760, obtain the issued claims from USPTO Patent Center (application 18/264,326), and commission a prior-art search against the 2021-02-18 priority date immediately — the art space (structured-light spot-pattern depth sensing plus NIR skin/material classification) is crowded with 2010s BASF/trinamiX publications and third-party face-unlock references that were never tested before the Board.
• Confirm the asserted-patent list. No public source enumerates the seven patents in trinamiX/BASF v. Apple. Pull the complaint from PACER / the W.D. Tex. Midland docket (possible case No. 7:26-cv-00348 per an unverified aggregator) to confirm whether 12,361,760 is asserted, which claims are identified, and which products are accused. Do not rely on news summaries for claim-level detail.
• If no IPR is filed, the absence will itself become a signal. Well-asserted, commercially valuable face-authentication patents of this kind typically attract IPR challenges within the first year of district-court enforcement. Monitor the PTAB trial-search database (https://developer.uspto.gov/ptab-api and the PTAB E2E system at https://ptab.uspto.gov) for any petition naming 12,361,760 or its family members, and monitor the W.D. Tex. docket for a motion to stay pending any future PTAB proceeding. As of today, there is no FWD, no panel, and no opinion to cite — the Board's docket for this patent is empty, and the litigation docket is barely 24 hours old.

I'll research the assignment chain for US 12361760, starting with the USPTO assignment records and cross-referencing litigation and entity status.

Inventors

All six named inventors assigned their interests to trinamiX GmbH (Ludwigshafen am Rhein, DE) in a single recorded "Assignment of Assignors' Interest" (recorded 2023-08-04; corrective recorded 2023-08-09):

• Peter Schillen
• Benjamin Guthier
• Friedrich Schick
• Manuel Guenther
• Lars Diesselberg
• Christian Lennartz

Employer at time of filing: trinamiX GmbH — all six are listed in the record as inventors on the PCT application (PCT/EP2022/053941, filed 2022-02-17) and the recorded assignment is the standard employment-style inventors-to-employer conveyance. Friedrich Schick's inventor profile confirms a long-running inventor relationship with trinamiX GmbH dating to 2016 (12 granted US patents assigned to trinamiX). No unusual pattern — no evidence any inventor departed the company or that the assignment was anything other than the routine employer capture on a PCT national-phase entry.

Original assignee

• Entity on the issued patent: TrinamiX GmbH (recorded per USPTO assignment and Google Patents legal events as "TRINAMIX GMBH"). The same entity is recorded as applicant/assignee throughout the family (US12288421B2, US12530925B2, US20240037201A1, US20260105781A1).
• Line of business: trinamiX GmbH is a technology company (reported as a BASF subsidiary) developing spectroscopy and 3D-sensing / material-classification technology, including face-authentication and presentation-attack-detection (PAD) components ("Hertzstueck™" infrared sensors are cited in the specification). The company develops and licenses component technology for mobile-device face unlock rather than selling end-user phones itself.
• Product embodiment: The claims describe a face-authentication method of the type trinamiX markets for integration into mobile devices (Face ID-style secure unlock); the specification names trinamiX's own commercial infrared sensor brand. Whether trinamiX "ships" a product embodying the claims itself, or only licenses the method to device makers, is not determinable from my sources.
• Current status: Operating. No bankruptcy, acquisition, or dissolution found. The reported September 2026 trinamiX/BASF district-court complaint against Apple (see below) is consistent with an active assertion posture by the operating company, not a distressed sale.

Assignment timeline

Important scope caveat: I could not directly query the USPTO Assignment Center (assignmentcenter.uspto.gov) in this session — the tool available to me was general web search only. The entries below come from the Google Patents legal-events/assignment feed for US12361760B2 (which mirrors USPTO assignment recordation data), including the corrective assignment's own citation of Reel 064497 / Frame 0952. Correspondent-of-record names are not published in the sources I could retrieve, so the correspondent field for each entry is unknown rather than absent. Any additional post-issuance records would need verification against the Assignment Center's full abstract of title.

• executed date not available / recorded 2023-08-04 — Reel 064497 / Frame 0952 (reel/frame cited by the later corrective record)

  • Conveyance: Assignment of Assignors' Interest
  • Assignor: Peter Schillen; Benjamin Guthier; Friedrich Schick; Manuel Guenther; Lars Diesselberg; Christian Lennartz
  • Assignee: TrinamiX GmbH
  • Correspondent: not available in sources consulted
  • Context: Routine pre-issuance capture of inventors' rights to employer on the PCT national-phase application; recorded ~18 months after the PCT filing date, consistent with national-phase entry timing.

• executed date not available / recorded 2023-08-09 — reel/frame not stated in the sources I retrieved (corrective record references prior record at 064497/0952)

  • Conveyance: Corrective Assignment (correcting the title to "Optical skin detection for face unlock"; the originally recorded cover sheet carried the typo "OPICAL SKIN SENTECTION FOR FACE UNLOCK")
  • Assignor: Peter Schillen; Benjamin Guthier; Friedrich Schick; Manuel Guenther; Lars Diesselberg; Christian Lennartz
  • Assignee: TrinamiX GmbH
  • Correspondent: not available in sources consulted
  • Context: Paperwork correction only — same parties, same interest; no change in ownership.

Post-issuance assignments: none found. The Google Patents legal-events record for US12361760B2 shows no transfer after grant (2025-07-15); the only later event is the 2025-12-16 priority claim to continuation US 19/420,836 (US20260105781A1), still naming trinamiX GmbH. The recorded chain is therefore: inventors → TrinamiX GmbH (2023), with trinamiX GmbH still the owner on issuance and as of the latest available records. If the USPTO Assignment Center shows no additional records on direct query, that is itself a finding: the original assignee still owns the patent.

Timeline diagram

timeline
    title Ownership of US 12361760
    2022 : PCT filed by TrinamiX GmbH
    2023 : Inventors assign to TrinamiX GmbH
         : Corrective assignment recorded
    2024 : US publication US20240037201A1
    2025 : Patent granted US12361760B2
    2025 : Continuation filed US19420836

NPE / troll-pattern signals

1. Shell-entity transfer — not present. The patent moved only from the six named inventors to their employer, TrinamiX GmbH, a German operating GmbH (BASF subsidiary per press reporting) — not to any "IP / Licensing / Holdings" LLC, registered-agent address, or single-purpose Delaware/Texas LLC. No evidence of shell-entity involvement anywhere in the recorded chain (Reel 064497/0952 and the 2023-08-09 corrective).

2. Known asserter in the chain — not present. TrinamiX GmbH is not on any known NPE list (Acacia, Marathon, IV, Wi-LAN, Conversant, Vringo, Pendrell, Round Rock, etc.). It is an operating technology company. No Unified Patents/RPX-listed high-frequency plaintiff appears in the chain.

3. Repeat correspondent across the chain — unclear. The two recorded entries are inventor→employer conveyances to the same assignee; prosecution for the family is handled by Armstrong Teasdale LLP (per family records), a full-service firm that also does operating-company work — a single firm appearance is not an NPE tell. I could not retrieve correspondent-of-record names from my sources; recurrence cannot be assessed. This signal should be rechecked against the Assignment Center PDFs.

4. Cascading transfers — not present. Exactly two records, both in August 2023, both to the same assignee, one being a ministerial corrective. No chained LLCs, no multi-hop transfers.

5. Pre-litigation transfer — not present. The only transfers (2023) predate the reported September 2026 trinamiX/BASF v. Apple complaint by roughly three years, and they are the ordinary inventors-to-employer conveyance, not an arrangement of the chain to enable assertion. The patent remained with the operating assignee through the alleged complaint date.

6. Bankruptcy fire-sale — not present. No bankruptcy proceedings involving TrinamiX GmbH or BASF were found; no trustee or auction sale appears in the chain.

7. Privateering — not present. No operating-company-to-NPE transfer exists; per press reporting trinamiX (with parent BASF) is itself the plaintiff in the reported Apple action — the inverse of privateering (operating company asserting directly). Note: an unverified court-aggregator record styles the plaintiff as "Trinamix Sensing LLC," which if accurate would warrant re-examination, but no assignment to any such LLC appears in the recorded chain and the aggregator is expressly unverified.

8. Defensive aggregator (anti-NPE) — not present. The chain does not terminate at RPX, AST, LOT, Unified Patents, or OIN. The patent remains with the original operating assignee; there is no defensive neutralization in the record.

Verdict

Insufficient data — with the important qualifier that every recorded datum points to clean operating-company ownership, not NPE activity.

The recorded chain (Reel 064497/0952 and the 2023-08-09 corrective, both to TrinamiX GmbH) shows only the standard inventors-to-employer conveyance, no post-issuance transfers, and no NPE indicia of any kind. The reason I cannot return "operating-company assertion" with confidence is purely evidentiary: I could not query the USPTO Assignment Center directly in this session (so post-issuance records and correspondent names remain unverified), and the reported trinamiX/BASF v. Apple action (W.D. Tex., ~Sept. 2026) does not enumerate the seven asserted patent numbers in any source I could retrieve, so I cannot confirm this specific patent is being asserted. If direct Assignment Center and docket checks confirm (a) no additional recorded transfers and (b) US 12,361,760 among the asserted patents in that suit, the correct verdict would be operating-company assertion — a BASF subsidiary enforcing its own face-unlock portfolio, with zero shell-entity, cascading-transfer, or repeat-correspondent signals.

Verification links:

• USPTO Patent Assignment Search (search patent number 12361760): https://assignmentcenter.uspto.gov/
• USPTO legacy assignment index: https://assignment.uspto.gov/patent/index.html
• Google Patents legal events for US12361760B2: https://patents.google.com/patent/US12361760/en

I’ve reached the tool-call limit for this session, so I’ll answer directly from (a) the authoritative full patent text supplied in your message, and (b) the USPTO/Google Patents/Justia search results captured above. I flag reliability limits where they exist rather than fabricate confirmations.

1. USPTO verification of the number (as requested)

• Searching USPTO sources for the specific number 12361760 returns US12361760B2 — “Optical skin detection for face unlock”, TrinamiX GmbH, granted July 15, 2025 (application US 18/264,326, filed Feb. 17, 2022; priority EP 21157800.0, Feb. 18, 2021; publication US20240037201A1). The Justia copy (patents.justia.com/patent/12361760) confirms the same patent and reproduces the claims, including claim 1 in full.
• No different/similar-numbered patent was substituted. Sibling applications US20240005703A1, US20240331450A1, US20250252783A1, and granted US12530925B2 carry the same title but are other family members, not patent 12361760.
• Caveat: I could not pull the USPTO “References Cited” front-page image for US12361760B2 itself. The citation lists below were surfaced from the same-family publication US20240005703A1 (which displayed “Citations (4)” and a family-level cite list of 8) and from the patent’s own specification text. The examiner’s formal citation list on the issued patent should be verified in USPTO Patent Center/PatFT if precision is required.

2. The granted claim set relevant to the § 102 analysis

Claim 1 (verbatim from Justia) is a four-step method: (a) face detection — capture a first 2D image, detect a face by identifying a predefined geometric face feature with a processing unit; (b) skin detection — project an illumination pattern of multiple illumination features, capture a second (reflection) image, analyze the beam profile of reflection feature(s) located in the second-image region corresponding to the first-image face region, derive a material property, and characterize the face as “skin” if the property matches a skin-characteristic property; (c) 3D detection — analyze beam profiles of at least four such reflection features, derive a depth level, and characterize the face as a 3D object if the depth level deviates from a predefined plane-object depth; (d) authentication — authenticate only if the face is both “skin” and “3D object.” Dependent features (from the specification’s embodiment list, exact dependent-claim text not separately retrieved) include machine-learning/CNN skin classification models, NIR illumination (≈940 nm), periodic laser-spot grids, depth-from-photon-ratio (DPR) quotients, and mobile-device/computer-program categories.

3. Citation list surfaced for the patent (family-level)

From the sibling publication US20240005703A1 (“Citations (4)”), all four are U.S. applications:

A family-level list of 8 additional references was also surfaced: WO2015054597A2 (Moffitt Cancer Center — texture-feature tumor imaging), US10368795B2 (Canfield — acne imaging), KR102452770B1 (trinamiX — optical detector, DPR family), US10922395B2 (STMicroelectronics — ToF facial authentication), US10990805B2 (Apple — hybrid-mode illumination for Face ID–style recognition), CN111164610B (Shenzhen Goodix — anti-spoof real-face sensing), JP2022525204A (trinamiX — detector identifying material properties, i.e., the WO 2020/187719 family), and CN114676403A (Huawei — authentication system). Because these came from the sibling page rather than the 12361760 front page, treat them as family-level citations, not confirmed front-page citations of 12361760.

4. Prior art embedded in the patent’s own specification (confirmed part of the 12361760 record)

The specification explicitly names and incorporates: US 2019/213309 A1 (face authentication with a ranging sensor using time-of-flight plus reflectance; in the “Prior Art” section); WO 2018/091649 A1, WO 2018/091638 A1, WO 2018/091640 A1 (trinamiX depth-from-photon-ratio/beam-profile-analysis); WO 2020/187719, EP 20159984.2 (filed Feb. 28, 2020) and EP 20 154 961.5 (filed Jan. 31, 2020) (material classification from coherent-light reflection/beam profiles); and EP 20 170 905.2 (filed Apr. 22, 2020, hexagonal laser-spot grid).

5. Which claims each reference potentially anticipates under § 102

For § 102(a)(1)/(a)(2) against an effective filing date of Feb. 18, 2021, a single reference must disclose every limitation of a claim, arranged as claimed. On the record available, no single located reference appears to anticipate claim 1, because each lacks at least one of the claim’s defining limitations — most commonly the requirement of classifying skin from a material property derived from beam-profile analysis of reflection features in the face-region of a second image, and/or the conjunctive requirement of authenticating only if both the skin test and a ≥4-feature 3D test pass.

• US20160242853A1 (Elwha): Potentially anticipates no claim. Unrelated technical field; no face detection/skin/3D/authentication combination disclosed.
• US20160335483A1 (Intel): Potentially anticipates no claim of the granted method as a single reference. It is relevant to liveness detection generally (element b’s anti-spoof goal) but does not disclose projecting a multi-feature illumination pattern and classifying skin from beam-profile material properties, nor the combined skin+3D authentication gate. Best used in a § 103 combination; may map to a broad reading of step (b) only if the beam-profile/material-property limitations were construed very expansively — which they should not be.
• US20210049391A1 (Qualcomm): Closest prior art on facial-liveness/spoof detection for face unlock. It is relevant to steps (b)–(d) conceptually, but (i) it does not disclose deriving a material property from beam-profile analysis of reflection features and characterizing skin thereby, and (ii) it publishes Feb. 18, 2021 — the same day as the EP priority date, so it is not “before” the effective filing date and would not be § 102(a)(1) prior art if the priority claim holds; it could be § 102(a)(2) prior art only if the effective filing date were deemed Feb. 17, 2022. No anticipation of claim 1.
• US20220222464A1 (Ford): Filed Jan. 13, 2021 (before the Feb. 18, 2021 priority date) and published Jul. 14, 2022, so it can qualify under § 102(a)(2). It is relevant to material classification/spectroscopy (element b’s material-property concept) but does not disclose face detection, geometric-feature identification, the four-reflection-feature 3D test, or skin+3D conjunctive authentication. No anticipation of claim 1; possible § 103 component.
• US 2019/213309 A1 (as written in the specification; i.e., a 2019 US publication of the form US2019/0213309A1): The specification’s own closest background art — face authentication using a ranging sensor with time-of-flight distance plus reflectance and processor-triggered facial recognition. It is the strongest candidate on the “reflectance/material + distance + authenticate” theme, but it lacks (i) detection of a face by identifying a predefined geometric feature in a first image, (ii) projection of a multi-feature illumination pattern and beam-profile/material-property skin classification, and (iii) use of ≥4 beam-profile-derived depths as a separate 3D-object test. Potentially anticipates no granted claim as a single reference; highly relevant under § 103 for steps (c)–(d).
• trinamiX’s own earlier filings (WO 2018/091649 A1 et al. — DPR depth from beam profiles; WO 2020/187719 / EP 20159984.2 / EP 20 154 961.5 — material classification from beam profiles): These disclose the core beam-profile-analysis tools (quotient Q, spot-shape/translucence/luminance/speckle filters, skin-vs-material classification) and could potentially anticipate dependent-feature limitations directed to DPR-based depth determination or beam-profile material filters if any claim is drawn to those alone. They do not anticipate claim 1, which requires the full face-authentication pipeline combining skin + 3D tests. Note also the incorporation-by-reference makes these the most technically probative references for claim construction.
• Family-level references (STMicro US10922395B2, Apple US10990805B2, Goodix CN111164610B, Huawei CN114676403A): All are face-authentication/anti-spoof systems using ToF, dot-projector illumination, or real-face sensing; each is relevant to steps (a)–(d) generically, but none located discloses skin classification via material property from reflection-feature beam profiles as claim 1 requires. No single-reference anticipation of claim 1 identified on the available information.

6. Bottom line

• The most relevant prior art for US 12361760 is the cluster of face-unlock spoof/liveness references (Qualcomm US20210049391A1, Intel US20160335483A1, STMicro US10922395B2, Apple US10990805B2, Goodix CN111164610B) combined with trinamiX’s own beam-profile material-classification and DPR filings (WO 2020/187719 family, WO 2018/091649 family) and the background reference US 2019/213309 A1.
• On the record available, no citation is a clean single-reference § 102 anticipation of claim 1; the realistic validity attack is § 103 obviousness combining a liveness/face-unlock reference (for steps a, c, d) with a beam-profile material-classification reference (for step b). The strongest § 102 concern would be against a narrow dependent claim directed only to the beam-profile material-classification or DPR sub-feature, where trinamiX’s own earlier publications are the primary risk.
• Verification caveat: Because I could not retrieve the USPTO “References Cited” page for the granted 12361760B2, confirm the examiner’s actual citation list in USPTO Patent Center before relying on this mapping; the four numbered citations above were displayed on the same-family publication US20240005703A1, and the family list came from that page’s family grouping.

Obviousness Analysis — U.S. Patent No. 12,361,760 (US12361760B2), "Optical skin detection for face unlock"

0. Scope, legal framework, and an important caveat on the claim set

Claim-reconstruction caveat (read this first). As flagged in the earlier sections of this analysis, I have not retrieved the verbatim text of the issued claims from USPTO Patent Center; the claim set analyzed here is reconstructed from the specification's Summary of Invention and the numbered "Embodiment" items (which track the intended claims 1–15). The analysis therefore maps prior art against the reconstructed independent method claim (face detection → skin detection via beam-profile-derived material property → 3D detection via beam-profile analysis of at least four reflection features → authentication gated on both) and its dependent features (Embodiments 2–15). If exact claim wording differs on any disputed limitation, the element-by-element maps below would need adjustment. The patent is classified in G06V40/40 (spoof detection/PAD), G06F21/32, G01S17/894, G06T7/50 — consistent with the reconstruction.

Statutory framework. The patent claims priority to EP 21157800.0 (filed 2021-02-18) via PCT/EP2022/053941 (filed 2022-02-17), making the effective filing date under AIA § 100(i)(1)(A) February 18, 2021. Prior art under § 102(a)(1) must have been publicly available before that date; § 102(a)(2) art must rest on an application effectively filed before that date. Under § 103, the question is whether the claimed subject matter as a whole would have been obvious to a person of ordinary skill at that time, under Graham and KSR — including whether a skilled artisan would have had reason to combine known elements with a reasonable expectation of success.

One self-inflicted complication that must be handled candidly. A large share of the closest technical disclosures — trinamiX's own "live skin"/Beam Profile Analysis work — was made by the patent owner or its related filings. Specifically:

• The WO 2018/091649 A1, WO 2018/091638 A1, WO 2018/091640 A1 depth-from-photon-ratio (DPR) family published in 2018, i.e., more than one year before the 2021-02-18 effective filing date. The § 102(b)(1) inventor-grace exception therefore does not rescue them (the exception only covers disclosures within one year of the effective filing date), and the § 102(b)(2)(C) common-ownership exception only applies to § 102(a)(2) "secret prior art," not to 2018 printed publications. These 2018 trinamiX publications are available prior art.
• By contrast, trinamiX's 2020 material/skin-classification filings (EP 20159984.2, filed 2020-02-28; EP 20 154 961.5, filed 2020-01-31, and their WO/US progeny such as WO 2020/187719 and US 2024/0027188 A1) and the February 19, 2020 / July 2020 "live skin" press and Qualcomm-blog disclosures fall inside the one-year window before the 2021-02-18 effective filing date and/or are commonly owned; they are likely excepted from use as § 102/§ 103 references under § 102(b)(1) and (b)(2)(C). I therefore treat them below not as primary references but as permissible evidence of the state of the art and of industry motivation (as discussed in § 5).

The combination case below is deliberately built primarily on third-party prior art that indisputably predates 2021-02-18, plus the 2018 trinamiX DPR family.

1. The prior-art record

A. US 2019/0213309 A1 — "US 2019/213309 A1" (admitted prior art in the patent's own background)

• What it is: The patent's own Prior Art section describes a "system and method of authenticating a user's face with a ranging sensor. The ranging sensor includes a time of flight sensor and a reflectance sensor. The ranging sensor transmits a signal that is reflected off of a user and received back at the ranging sensor. The received signal can be used to determine distance between the user and the sensor, and the reflectance value of the user. With the distance or the reflectivity, a processor can activate a facial recognition process in response to the distance and the reflectivity."
• Significance: This single reference already teaches the conceptual spine of the claimed method: face authentication gated on both a distance/3D measurement and a reflectance/material-ish measurement, with recognition activated only if the sensed quantities are consistent with a live user. My independent verification search for this document returned no result, so I rely on the patent's own characterization — but for § 103 purposes, an admission in the specification is a permissible and potent starting point.
• Date: Published (per its publication number) in 2019, before the critical date.

B. Koninklijke Philips N.V. — US 10,242,278 B2 / WO 2016/116307 family (device and method for skin detection)

• What it is: "A device and a corresponding method for skin detection to enable a reliable, accurate and fast detection ... compris[ing] an illumination unit configured to project a predetermined illumination pattern onto a scene, an imaging unit configured to acquire an image of the scene, and an evaluation unit configured to evaluate the acquired image by analyzing the images of illumination patterns as reproduced in the image and to detect skin areas within the image and distinguish them from non-skin areas within the image based on said analysis." (US 10,242,278 B2 abstract, filed 2015-11-24, granted 2019-03-26; Philips; inventors Makkapati and Kirenko; main CPC G06K9/2036 — i.e., pattern-based image sensing.)
• Also relevant: The related disclosure US 2016/0155006 A1 is cited in the trinamiX material-detection family (EP 3 938 802 B1, background at [0003]) as describing the same Philips-type skin detection via analysis of a reproduced illumination pattern.
• Significance: Teaches, in the biometric/PAD context, projecting a patterned illumination onto a scene and classifying skin vs. non-skin from how the pattern is reproduced — the core of claimed step (b) — years before the priority date.
• Sources: https://www.expertkg.com/patent/?patent_id=[10242278](/patent/10242278) ; https://data.epo.org/publication-server/rest/v1.2/publication-dates/2025-05-21/patents/EP3938802NWB1/document.pdf

C. Microsoft Technology Licensing — US 9,947,099 B2 ("Reflectivity map estimate from dot based structured light systems")

• What it is: Granted 2018-04-17 (filed 2016-07-27). Systems/methods for determining both a depth map and a reflectivity map from a single structured-light dot image: project a plurality of dots, capture the image, determine depth by triangulation, then "detect one or more projected dots," analyze "the intensity across the size and/or shape of the dot," correct for distance/angle/illumination, and generate a per-dot reflectivity/intensity profile.
• Significance: Teaches the exact hardware geometry of the claimed method (a dot-pattern projector + camera producing a reflection image whose individual spots each have an analyzable beam profile/intensity distribution), and — critically — that the same captured dot pattern can be processed twice: once for per-spot depth (3D) and once for per-spot physical/reflectance properties. It is the closest third-party analogue to the "one camera frame yields both the skin features and the 3D features" architecture of the claim.
• Sources: https://patents.google.com/patent/[US9947099B2](/patent/US9947099B2)/en ; https://patentimages.storage.googleapis.com/89/2a/29/39fd320529d1d9/US9947099.pdf

D. trinamiX DPR family — WO 2018/091649 A1, WO 2018/091638 A1, WO 2018/091640 A1 (2018)

• What it is: The depth-from-photon-ratio ("beam profile analysis") lineage incorporated by reference into the patent itself. Teaches determining, for each reflection feature (laser spot) in a reflection image, a longitudinal coordinate from the shape of its beam profile — e.g., by integrating a first area (edge information) and a second area (center information) of the spot, forming a quotient Q, and mapping Q to depth via a stored relationship.
• Significance: Provides the per-spot, beam-profile-based 3D detection of claimed step (c) in detail, and confirms that persons skilled in the art treated spot beam profiles as information carriers (depth and material-dependent shape). Because these published in 2018, they are available prior art notwithstanding common ownership (see § 0).
• Sources: Referenced throughout the patent's own specification (e.g., DPR technique paragraphs) and confirmed in the search results (e.g., US20240284031A1, KR20220147604A, US20240013416A1, each incorporating them).

E. State of the art / motivation evidence (not primary § 102 references)

• trinamiX/Qualcomm announcements of February 19, 2020 and July 2020 describing "Beam Profile Analysis," which "extracts 3 data streams from a single camera system: a 2D IR image, a 3D depth map and most uniquely — material classification," explicitly to block "a mask, a high-resolution print or even a 3D rendering" in smartphone face unlock (PR Newswire; Qualcomm "OnQ" blog).
• The patent's own Background (PRIOR ART section) documenting the known attack landscape: 2D photos, video replays, printed/tablet presentations, and "high quality masks generated, for example, by 3D printing, careful 3D-arrangement of 2D photos, or hand-crafted silicone or latex masks," and the known countermeasure sequence (NIR cameras to exclude displays; 3D cameras to exclude planar photos).
• As explained in § 0, these items are likely excepted as prior art under § 102(b)(1)/(b)(2)(C) against this application, but they are admissible and highly probative on (i) the level of ordinary skill, (ii) the recognized problem, and (iii) the existence of an industry-wide motivation to combine 2D + 3D + material/skin sensing in one face-unlock pipeline.

2. Level of ordinary skill (POSITA)

A POSITA for this art would have, at a minimum: a B.S. (more typically an M.S.) in electrical engineering, computer science, physics, or optical engineering; 2–4 years' experience in at least one of (i) biometric authentication and presentation-attack detection, (ii) structured-light / active 3D sensing, or (iii) optical material characterization; working familiarity with NIR CMOS cameras, VCSEL/laser dot projectors and diffractive optical elements, 2D face detection and landmark CNNs, and image-based classification. The specification's own cross-references (DPR lineage, material-filter lineage, face-recognition surveys) confirm that the inventor-side view of the field treats these as mature, inter-changeable building blocks.

3. The obviousness case — combination of References A–D

3.1 Primary combination: US 2019/0213309 A1 (A) + US 10,242,278 B2 / WO 2016/116307 (B) + US 9,947,099 B2 (C), optionally + WO 2018/091649 et al. (D)

Claim 1 (reconstructed) — element-by-element map

Why a POSITA would combine A + B + C (± D):

1. Same field, same problem. Each reference addresses secure face/scene authentication and the known spoof set (prints, screens, masks). The patent's own Prior Art section concedes that (i) NIR cameras, (ii) 3D cameras, and (iii) skin/material classifiers were each known, piecemeal, countermeasures — the claimed advance is only their integration.
2. Hardware compatibility and efficiency. (C) demonstrates that a single structured-light dot projector and camera yield both depth and per-dot reflectivity from one frame. (B) demonstrates pattern-based skin detection. The skilled artisan seeking to build a PAD system that rejects both flat media and masks would immediately see that (B)'s classifier can consume (C)'s per-dot profiles — no new sensor needed. (D) shows that beam-profile analysis itself is a single tool that outputs both depth and material cues from the same spots, making the "same frame → skin + 3D" design the natural, even inevitable, architecture.
3. Known technique applied to known problem. Under KSR, combining a known skin-classification technique (B) with a known dot-profile/depth technique (C/D) to solve the acknowledged mask/photo attack problem is the paradigm of "the combination of familiar elements according to known methods," yielding a predictable result: an anti-spoof gate. The conjunctive "authenticate only if both" logic is supplied directly by (A).
4. Industry pressure confirms motivation. The 2020 trinamiX/Qualcomm disclosures (and the commercial race around Apple Face ID-class dot-projector unlock) show multiple actors were motivated to fuse 2D + 3D + material sensing in a single face-unlock camera — objective evidence of a recognized problem and a design trend, even if those specific disclosures are not themselves usable as references.

3.2 Alternative combination: Philips (B) + trinamiX DPR (D) as the depth leg, without reliance on Microsoft (C)

If a challenger wants a combination that tracks the patent's own preferred implementation more literally: (B) supplies skin detection from an illumination pattern; (D) supplies per-spot depth from beam-profile analysis (quotient Q of edge/center integrals); a POSITA would combine them because (D) is expressly a material- and depth-sensitive analysis of the same laser spots (B) classifies — and the specification itself describes the two claimed analyses as performed on "the exact same" reflection features and frame. The 2018 publication of (D) makes it usable. The only missing element is the face-detection front-end and the conjunctive gating, both supplied by (A)/conventional face detection as above.

3.3 Dependent features (reconstructed claims 2–11 and the Embodiment list)

• Enrollment/identification (Embodiment 3 / claim 3): Storing a 2D image of the user at enrollment and comparing it with the first image is textbook face-recognition practice (the spec itself describes it as conventional). Adding 3D/material "fingerprints" to an enrollment template is the obvious extension of using the same sensors for recognition and PAD.
• Parametrized skin-classification model via machine learning; CNN input = second image (Embodiments 4–5 / claims 4–5): Using CNNs to classify image patches — including skin/no-skin from reflection images — was routine by 2021; the specification's own material-filter discussion concedes feature-based methods "may be used in combination with machine learning." Training a classifier with an optimization target is the definition of supervised learning. (C)'s per-dot features and (B)'s skin labels give the obvious training data.
• 2D face + landmark detection feeding region-specific skin models (Embodiment 6 / claim 6): Landmark detectors were standard (see spec's own citation of face-recognition surveys); training separate classifiers for forehead vs. cheek regions is routine data partitioning, driven by the known physiological fact (acknowledged in the spec) that skin optics vary by facial zone.
• Periodic grid of laser spots (Embodiment 7 / claim 7); hexagonal/periodic tilings: Dot grids and diffractive-optics spot projectors are the standard structured-light pattern (Microsoft's own Kinect/AR projector lineage; (C)'s dots; (B)'s predetermined pattern). A periodic grid is the obvious default.
• NIR illumination, in particular ~940 nm (Embodiments 8–9 / claims 8–9); NIR camera (Embodiment 11 / claim 11): NIR active illumination for face unlock was standard (to defeat display attacks — the patent's Background admits this). 940 nm is a commodity VCSEL/LED wavelength chosen precisely because (per the spec) "Melanin absorption runs out so that dark and light complexion reflect light almost identically" — an express, routine wavelength-selection rationale; the spec itself lists five alternative wavelengths.
• Plurality of second images; reflection features from multiple frames used for skin and/or 3D (Embodiment 10 / claim 10): Frame-averaging/temporal sampling for classification stability is conventional (the spec frames it as a "more stable classification" option).
• Computer program / storage medium / mobile device / use for PAD (Embodiments 12–15 / claims 12–15): These categories track the method; the mobile-device claim recites only the standard hardware (camera, illumination unit, processing unit) that (B)/(C)/(D) each describe, configured to execute the obvious method. Statutory-category claims of this type rise and fall with the method claim.

4. Strongest and weakest points of the obviousness case

Strongest for the challenger

• The prior art is overwhelmingly crowded, and the "problem" is admitted. The specification concedes each ingredient (2D face detection, NIR to defeat displays, 3D to defeat planar photos, skin/material classification to defeat masks) was known, and that known PAD systems "require several video frames," motivating a single-frame fused solution. Under KSR, an "obvious to try" with a finite, known set of building blocks and a predictable architecture is textbook obviousness.
• (A) supplies the conjunctive gating logic directly (activate recognition only on distance + reflectivity), so the "authenticate only if skin AND 3D" limitation is not novel in structure.
• (C) supplies the "one dot image → per-spot physical property + depth" architecture, which is the heart of the claimed efficiency benefit.
• Numerical/geometric limitations ("at least four," "image region corresponding to the face region," "depth level of plane objects") are routine implementation choices.

Weakest for the challenger (arguments the patentee would make)

• No single reference performs skin classification from per-spot beam-profile material properties in a face-authentication pipeline. (B) classifies skin from pattern reproduction but does not expressly compute a per-spot "material property" such as scattering/translucency; (C) computes reflectivity but does not classify skin. A patentee will argue the combination requires hindsight and that the specific insight — that volume-scattering skin produces a characteristic NIR spot profile distinct from silicone/latex/foam — is not fairly suggested by (B)+(C) alone. The trinamiX 2020 material-classification filings that do teach this are (likely) unavailable as references due to the § 102(b)(1)/(b)(2)(C) exceptions, leaving a genuine gap a challenger must bridge with (D)'s material-sensitive dicta plus expert testimony on the state of the art.
• Secondary considerations are not yet developed (no litigated validity record as of the April 2026 current date; the PTAB docket is empty). If the patentee can show long-felt need, industry adoption (e.g., licensing/implementation of the BPA approach), or unexpected results (ethnic-origin agnosticism at 940 nm), those would cut against obviousness.
• Priority/grace complexities: If any challenged claim is shown to be entitled to a filing date later than 2021-02-18 (e.g., if certain dependent features lack support in EP 21157800.0), additional 2020–2021 trinamiX publications could become available — which would strengthen the challenger's case considerably. Conversely, if the examiner's file history shows the skin-material analysis was the point of allowance, the challenger should expect the fight to concentrate there.

5. Bottom line

Under § 103, the reconstructed independent method claim and its dependent features face a substantial obviousness challenge on this record. The most powerful combination is:

US 2019/0213309 A1 (face authentication gated on distance + reflectance) in view of Philips US 10,242,278 B2 / WO 2016/116307 (illumination-pattern skin/non-skin detection for biometrics) and Microsoft US 9,947,099 B2 (single structured-light dot frame yielding both a per-dot depth map and per-dot beam-profile/reflectivity analysis), optionally further in view of the 2018 trinamiX DPR family WO 2018/091649 A1 / WO 2018/091638 A1 / WO 2018/091640 A1 (per-spot beam-profile depth and material-sensitive analysis), with the motivation supplied by the admitted mask/photo/display spoof problem and the industry-wide push (evidenced by 2020 trinamiX/Qualcomm disclosures) toward fusing 2D + 3D + skin sensing in one face-unlock camera.

The contested core will be whether "skin characterization from a per-spot beam-profile material property" is fairly suggested by the Philips/Microsoft/DPR combination — a fact-intensive question on which the usability (or not) of trinamiX's own 2020 material-classification filings and any objective indicia will likely decide the outcome. Because the PTAB record is empty and the first known assertion (trinamiX/BASF v. Apple, W.D. Tex., ~Sept. 2026 per prior research) is recent, no tribunal has yet tested these arguments; a defendant with standing should commission a claim-specific search and expert declaration on the skin-material element before the § 315(b) IPR window considerations noted in the PTAB section come into play.

Caveats restated: (1) this analysis is based on the reconstructed claim set, not the verified issued text; (2) US 2019/0213309 A1 could not be independently retrieved in my searches and is characterized from the patent's own Prior Art section; (3) the availability of trinamiX's 2020-era disclosures as references turns on priority-chain and § 102(b)(1)/(b)(2)(C) facts that should be verified against the file history.