US 10990805

I'll search for the specific patent number and any related 2026 litigation activity.

Let me run additional targeted searches on litigation and claim details.

Let me try to retrieve the complete claim set for the granted patent.

Summary — U.S. Patent No. 10,990,805

Caveat on method: I do not have a direct PATENTSCOPE/PatentCenter/CourtListener query interface here — only general web search. The bibliographic data and claims below are drawn from the Google Patents page for US10990805B2 and the Justia reproduction of the granted claims, cross-checked against the Google Patents family data already supplied in the prompt. Where I could not verify something, I say so.

Bibliographic data

Abstract (as issued; verbatim substance)

A facial recognition authentication process may utilize images of a user's face captured while the user is illuminated using both flood infrared illumination and patterned illumination (e.g., speckle pattern illumination). Because the face is illuminated by both, the captured images may contain both flood IR illumination data and depth map image data. Flood IR data may be generated from the images to assess two-dimensional features; depth map data may be generated from the pattern data in the images to assess three-dimensional (depth) features. The flood IR illumination data and the depth map image data may be used separately by the facial recognition authentication process to attempt to authenticate the user as an authorized user of the device.

Independent claims — plain language

Claim 1 (method) — the only independent method claim:

1. Receive an unlock request on a user interface of a device having a processor and memory.
2. In response, illuminate the user with both flood IR and patterned IR illumination.
3. Capture at least two images of the user with the device camera while the user is under both illuminations.
4. Generate flood IR image data from the at least two images.
5. Combine the at least two images into a composite image.
6. Generate depth map image data from the composite image (not from a single frame).
7. Authorize the user to perform an operation requiring authentication using a facial recognition process operating on both the flood IR data and the depth map data.

Note the narrowing versus the pre-grant publication: the issued claim 1 recites "at least two images" plus the express composite-image limitation, whereas claim 1 of US 2020/0082155 A1 recited "at least one image" with no composite step (the composite step appeared only in published claim 8). Prosecution thus folded the multi-frame composite feature into independent claim 1.

Claim 9 (device) — the only independent apparatus claim:
A device comprising a processor, memory, camera, user interface, at least one illuminator providing both flood IR and patterned IR illumination, and circuitry configured to: receive an unlock request; illuminate the user with both illuminations; capture a series of images in which at least some images are captured under both illuminations and at least one image is captured while the user is not illuminated (i.e., the "empty frame" concept of FIGS. 9–10); generate flood IR image data and depth map image data from the captured images; and authorize the user using a facial recognition process on that data.

Third independent claim — uncertainty. The Justia listing of the granted patent is truncated after claim 14 ("The device of claim 9, wherein the captured images comprise images of a face of the user…"). I could not confirm from the available sources whether the granted patent ends at claim 14 or contains a further independent claim (e.g., a computer-readable-medium claim) at claim 15+. The pre-grant publication had 15 claims with two independents (method claim 1, device claim 10); the granted set is renumbered (method claim 1, device claim 9), so the total count differs. Do not treat a claim-15 independent claim as confirmed.

Notable dependent claims

• 2/13 – flood IR data taken from image areas between pattern features (speckles/dots); parsing flood data from depth data.
• 3 – depth data obtained by assessing the projected pattern on the user.
• 4 – flood data generated separately from depth data (the two-path A/B architecture of FIG. 4).
• 6–7 – dual-path scoring: encode flood → flood feature vector; encode depth → depth feature vector; compare each to its own template space for a first and second matching score; derive a third score (e.g., average) and unlock if above threshold.
• 8 – simultaneous illumination during capture.
• 10–12 – first (flood) and second (patterned) illuminators, or a single illuminator providing both; user interface = the display.

2026 litigation / CAFC check

• No CAFC 2026 docket and no IPR/appeal docket involving US 10,990,805 was found in my searches. I cannot verify a negative exhaustively without Docket Navigator/CourtListener API access — treat this as "nothing found," not "nothing exists."
• The most prominent 2026 Face ID dispute is trinamiX Sensing LLC et al. v. Apple, Inc., No. 7:26-cv-00348 (W.D. Tex., filed Sept. 3, 2026), a district court case (not CAFC) asserting US 12,530,925; 12,288,421; 12,456,334; 12,361,760; 12,298,394; 12,332,352; and 12,461,238. US 10,990,805 is not among the asserted patents, and the complaint distinguishes Apple's original 2017 Face ID (which used Apple's own technology) from later material/skin-detection features.
• Date inconsistency to flag: your prompt states the current date is April 26, 2026, but my search results include content dated up to late September 2026 (system date 2026-09-28). If the April date controls, the trinamiX filing has not yet occurred. I have reported it because it is the only 2026 Face ID patent enforcement activity my searches surfaced, but it is not tied to this patent in any way I could document.

Bottom line

US 10,990,805 is an Apple Inc. patent (Kalscheur and Susskind) covering hybrid-mode illumination for Face ID–style authentication, in which a single capture session uses flood + patterned IR simultaneously so that one image set yields both 2D texture data and 3D depth data. The granted independent claims are narrower than the published application: claim 1 requires multiple images combined into a composite used for the depth map, and claim 9 claims a device that deliberately captures at least one un-illuminated ("empty") frame within the image series (the anti-spoofing/replay feature of FIGS. 9–10). I found no 2026 Federal Circuit activity for this patent.

I'll research litigation involving US patent 10990805. Let me search multiple sources.

Let me search patent litigation-specific databases with more targeted queries.

Litigation Search Report: US Patent 10,990,805

Executive Summary

No litigation, PTAB proceeding, ITC action, or Federal Circuit appeal involving U.S. Patent No. 10,990,805 ("the '805 patent") was found in my searches. Based on the sources available to me, this patent does not appear to have been asserted in any lawsuit, nor has it been challenged in any post-grant proceeding.

Patent Identification (verified)

Per the authoritative patent text and Google Patents record at https://patents.google.com/patent/[US10990805](/patent/US10990805)/en:

I confirmed this is the correct patent — an Apple patent on hybrid flood-IR + patterned-illumination facial recognition (Face ID-related). No similar-number results (e.g., other "…90805" patents or SEC filing ID "10990805" for Community Financial System, Inc.) were treated as relevant.

Litigation Found: None

My searches of litigation-covering sources (Federal Circuit opinions, Docket Alarm, RPX Insight/CourtListener, PACER-derived dockets, Unified Patents materials) returned no case in which US 10,990,805 is:

• asserted as a patent-in-suit by Apple (or any assignee), or
• challenged by a defendant via declaratory judgment, IPR, PGR, EPR, or ITC complaint.

No plaintiff/defendant pair, jurisdiction, case number, or filing date can be reported because no such case was located.

Important Distinction: BASF / trinamiX v. Apple (NOT this patent)

My search surfaced a highly relevant but distinct Face ID patent dispute that should not be confused with litigation over the '805 patent:

• Case: trinamiX GmbH and trinamiX Sensing LLC v. Apple Inc.
• Court: U.S. District Court for the Western District of Texas (Midland Division)
• Case No.: 7:26-cv-00348 (assigned to Judge David Counts)
• Filed: September 3, 2026
• Asserting party: trinamiX (a BASF subsidiary); Defendant: Apple Inc.
• Patents asserted: seven trinamiX patents (skin/material detection for anti-spoofing) — not the '805 patent, and not any Apple patent.
• Accused products: iPhone 15/16/17 families, iPhone Air, certain iPad Pro models.
• Status: Pending; trinamiX seeks damages, injunction, fees, and a jury trial. Apple had not publicly responded as of the reports.

Notably, this case is adverse in posture to the '805 patent: per a MacRumors discussion and associated analysis (https://www.macrumors.com/2026/09/08/apple-sued-over-face-id-patent-infringement/), Apple's US 10,990,805 is cited as prior art by the later trinamiX patent family — i.e., the '805 patent is being used as a defensive/prior-art reference, not asserted in litigation.

Confidence, Caveats, and Limitations

• Confidence the '805 patent has not been litigated: moderate-to-high, but not absolute. Apple owns the patent and generally does not assert its own consumer-device patents offensively, which is consistent with finding no suits. However, my review relied on web-search-accessible aggregators (Google Patents, Docket Alarm, RPX Insight, CourtListener, Unified Patents, Justia), not a direct, exhaustive PACER/Docket Navigator/Unified Patents caselist query for "10990805." A docket-level litigation search is the authoritative method and I could not complete a full enumeration before reaching my search limit.
• "Litigation" scope used: district court infringement/DJ suits, ITC § 337 actions, PTAB post-grant proceedings (IPR/PGR/EPR), and CAFC appeals. No results in any of these categories.
• No auto-correction applied: I interpreted "10990805" literally and excluded unrelated hits (e.g., a 2004 SEC Form 8-K bearing ID "10990805," and other "908"-suffixed patents such as Neo Wireless's US 10,833,908).

Recommended Next Steps for a Definitive Answer

To confirm the negative result conclusively, I recommend querying the following directly for "10,990,805":

1. Unified Patents Litigation Portal — https://portal.unifiedpatents.com/litigation/caselist
2. Docket Navigator / Lex Machina (full PACER docket search by patent number)
3. USPTO Patent Trial and Appeal Board (PTAB) proceeding search by patent number
4. USITC EDIS (for any § 337 investigation)
5. CAFC docket search for the patent number as a case document identifier

If those return nothing, the conclusion that US 10,990,805 has no known litigation would be firmly established as of the current date.

I'll verify the canonical (empty) proceedings list against live sources before reporting.

Proceedings overview

Total AIA trial proceedings on file for US 10,990,805: ZERO. The canonical structured block — sourced from the USPTO Open Data Portal (ODP) — returns no inter partes review, post-grant review, or covered business method proceeding for this patent, and my independent web searches (PTAB/PTABTS petition documents, Federal Circuit materials via CourtListener, Docket Alarm PTAB case pages, and general litigation-press searches) surfaced nothing that contradicts that negative result. There is therefore no breakdown to report: 0 active, 0 claims invalidated, 0 claims sustained, 0 settled, 0 institution denials. No proceeding sections follow, because there is no proceeding to describe — and I will not invent a docket number to fill the space.

Bottom-line defensive posture: this is the opposite of a "hardened by IPR" patent. No challenger has ever put these claims in front of the Board, so every claim is UNTESTED, the Patent Owner has incurred zero PTAB attrition, and no § 315(e) estoppel attaches to anyone — meaning a defendant today has a clean, unencumbered runway to file an IPR (subject only to the § 315(b) one-year bar if it has been served with an infringement complaint, and to § 325(d) discretion regarding art already considered at prosecution). Conversely, the absence of any IPR also means there is no publicly vetted invalidity roadmap you can harvest; you would be paying for the first pass at these claims.

---

Per-proceeding detail

None. There is no IPR####-#####, PGR####-#####, or CBM####-##### proceeding to summarize. Specifically:

• Type: n/a
• Filed: n/a
• Status: n/a — the ODP ingest reports no AIA trial record
• Judge panel: n/a — no panel has ever been assigned
• Petition grounds: n/a
• Institution decision: n/a
• Final Written Decision: n/a — no FWD exists, so there is no claim-level verdict and no panel reasoning to quote. Any claim that "claims X–Y were canceled" would be a fabrication.
• Settlement / termination: n/a
• Appeal: n/a — no FWD, hence no CAFC appeal docket
• Defensive value: The defensive value of the absence is that you face no estoppel, no adverse claim constructions from a prior panel, and no settled record. Any IPR you file will be the first, with the Board writing on a blank slate.

---

Strategic summary

Cancelled vs. sustained vs. untested. Because no AIA trial has ever been instituted, 100% of the claims of US 10,990,805 are UNTESTED. There are no canceled claims, no surviving-after-IPR claims, and no substitute claims from a motion to amend. I want to be explicit about a limitation of this record: the authoritative patent text supplied in the prompt is truncated before the claims section (it cuts off mid-sentence at the end of the FIG. 4 discussion of process 250), so I cannot state the exact number or wording of the claims from the material in front of me, and I will not guess. If claim-level precision matters for a demand-letter analysis, pull the claim text directly from https://patents.google.com/patent/US10990805/en (the "Claims" section) or the granted-PDF at https://patentimages.storage.googleapis.com/0c/11/33/41ce2f726b0e8e/US10990805.pdf.

Estoppel landscape. There is no § 315(e) estoppel in play, from any party, because no IPR/PGR has ever reached a final written decision on this patent. That is a material advantage for a would-be petitioner: you are not boxed out of any ground, you are not constrained by any prior petitioner's claim-construction positions, and you carry no privity taint from an earlier challenger. The only statutory gates you face are the ordinary ones — § 315(b) (one-year bar running from service of a complaint alleging infringement of this patent) and § 325(d) (discretionary denial where the same art or arguments were previously presented to the Office — relevant here because the '805 patent has a rich prosecution history, including a provisional priority claim to 62/730,115 and a CN family member CN 110895689 B).

Pattern signals. No petitioner has filed even one petition, so the classic red flags (serial filers, Unified Patents as a defensive aggregator, General Plastic-style multiple-petition gamesmanship) are absent by definition. No patent-owner appeal activity exists because there is nothing to appeal. This is consistent with the broader picture from the litigation section of this analysis: Apple, the assignee, is a defensive holder that has not asserted the '805 patent, so no accused infringer has yet had the incentive to fund an IPR against it. Note the related-family angle: the '805 patent shares its disclosure with Apple applications that published as US 2019/0042835 ("Multiple enrollments in facial recognition"), US 2019/0370449 ("Automatic retries for facial recognition"), and the granted US 11,151,235 ("Biometric authentication techniques," describing flood illuminator 430 / point illuminator 440 and pseudo-random capture-mode verification). Those siblings have likewise not surfaced as AIA trial subjects in my searches — but a challenger weighing an IPR on the '805 should consider the whole family, since a win on one member may not clear the portfolio.

Caveat on the negative. My confidence is "no proceedings found," not "proven zero." I could not execute a direct PTAB E2E / Docket Navigator / Lex Machina enumeration in this session; I relied on the ODP-sourced structured block plus web-search-visible aggregators. The negative is corroborated but not independently docket-verified.

---

Recommended next steps

Because there is no PTAB activity, say so plainly and act accordingly:

1. Confirm the negative on primary sources before relying on it. Query the PTAB End-to-End system by patent number at https://e2e.uspto.gov (Patent Number search, "10,990,805"), and cross-check the USPTO PTAB Decisions repository at https://www.uspto.gov/patents/ptab/decisions. Also check the Unified Patents litigation/PTAB portal (https://portal.unifiedpatents.com) and Docket Alarm's PTAB coverage. If all four return nothing, the zero-count conclusion is firm as of 2026-09-28.

2. Treat the patent as fully open to challenge — but move fast if you've been served. With no prior petitioner, you have no estoppel to fear, but the § 315(b) clock is personal to you: if you have received a complaint alleging infringement of the '805 patent, the petition must be on file within one year of service. There is no FWD to link and no disposition to quote, because none exists.

3. Do not expect trial-stage milestones. There is no institution-decision deadline running, no oral hearing scheduled, and no statutory one-year FWD deadline to track, because there is no trial. If that changes, the milestones would be: institution decision due ~6 months from petition filing, FWD due within 12 months of institution (§ 316(a)(11)), with the hearing typically ~2–3 months before the FWD.

4. Also rule out non-AIA Office proceedings. My review targeted AIA trials; a concurrent ex parte reexamination or reissue could exist outside that scope. A quick USPTO Patent Center "Transaction History" check for US 16/277,485 will close that gap. I have not verified this either way.

Sources consulted (all negative): https://patents.google.com/patent/US10990805/en (family and status), https://patentimages.storage.googleapis.com/0c/11/33/41ce2f726b0e8e/US10990805.pdf (granted text), https://ptacts.uspto.gov (PTABTS petition-document search — no hits for this patent), and the ODP-derived structured block reproduced in this prompt. No PTAB decision, institution decision, or Federal Circuit opinion citing or involving US 10,990,805 was located.

I'll research the assignment record for US 10,990,805 before drafting.

Let me try to retrieve the specific reel/frame for the Apple assignment.

Let me try to pin down the specific reel/frame and correspondent for this assignment.

I was unable to retrieve the reel/frame number or correspondent of record for the single Apple assignment before hitting my search limit. I will report exactly what is documented and flag the gaps rather than fabricate. Note the carried-forward date discrepancy (header says April 26, 2026; system date is 2026-09-28) — it does not affect this analysis.

---

Inventors

Two named inventors on US 10,990,805:

Unusual patterns — none found.

• No mass-departure signal. I found no evidence that either inventor left Apple within 12 months of filing, and no evidence of a portfolio fire-sale. The patent remains Apple-owned (see timeline). A single-inventor-pair, company-assigned, still-owned patent is the ordinary operating-company profile.
• No third-party co-owners. Both inventors assigned their entire interest to Apple; there is no university, government, or joint-development co-owner to fracture the chain.
• Caveat: inventor employer status and departure dates are not exposed by the assignment record, and I could not run a personnel/employment check. Treat "no departure signal" as "nothing found," not "verified absent."

---

Original assignee

Apple Inc. (Cupertino, California) — the entity named on the issued patent.

• Product embodying the claims: The '805 patent is directed to hybrid flood-IR + patterned-IR illumination for facial recognition authentication — i.e., Face ID. Apple shipped Face ID starting with iPhone X (2017) and has deployed a "hybrid" capture scheme in the TrueDepth camera system across iPhone/iPad Pro families. The patent's own specification ties the technique to a "mobile device" with camera, ISP, and secure enclave processor (SEP 112) — a direct description of Apple's Face ID hardware. I treat product embodiment as high-confidence but inferred from subject matter and the specification, not from an Apple admission that this specific patent covers a specific shipping product.
• Primary line of business: consumer electronics, software, and services (design and sale of iPhone, iPad, Mac, wearables, and related services).
• Current status: Operating — one of the largest public companies in the world; not acquired, dissolved, or in bankruptcy. (Consistent with the assignee-of-record being Apple throughout.)

---

Assignment timeline

One meaningfully recorded conveyance exists. Google Patents' legal-events block for US 10,990,805 shows exactly one assignment event on 2019-02-15 — inventors → Apple Inc. — and nothing afterward.

• Executed ~2019-02-15 / recorded 2019-02-15 — Reel not retrieved (see gap note)
  • Conveyance: Assignment (of assignors' interest)
  • Assignor: Micah P. Kalscheur; Joshua M. Susskind
  • Assignee: APPLE INC., One Apple Park Way / (historically) 1 Infinite Loop, Cupertino, CA 95014
  • Correspondent: Not retrieved. I could not obtain the recording attorney/firm of record for this reel from the sources available to me. I am not naming a correspondent because I did not verify one.
  • Context: Ordinary employment/invention assignment — the founders assigned their rights to their employer. This is the routine "inventor → operating company" capture, recorded on the same day the non-provisional application (16/277,485) was filed. It is not a sale, securitization, reorg, or transfer-to-asserter.

Gap note (important): The Google Patents legal-events entry reads verbatim: "2019-02-15 Assigned to APPLE INC. … APPLE INC. ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS). Assignors: KALSCHEUR, MICAH P., SUSSKIND, JOSHUA M." — it does not expose the reel/frame or correspondent. I attempted to pull the reel/frame from USPTO legacy-assignments and third-party mirrors and failed before my search limit. Two possibilities remain open and I cannot distinguish them from here:

1. The reel/frame exists but is simply not surfaced in the sources I could reach; or
2. The record retrieved is the "assignment of interest" event and a second recordation (e.g., a confirmatory assignment) may exist that I did not see.

Everything downstream of 2019-02-15 is a negative: no security agreement, no merger, no change of name (note: Apple Inc. did have a historical "Apple Computer, Inc. → Apple Inc." change of name in 2007, but that predates this patent and would not appear on this chain), no license, no release, and no reassignment. In plain terms: the patent has never changed hands since the inventors assigned it to Apple.

---

Timeline diagram

timeline
    title Ownership of US 10990805
    2018 : Provisional 62730115 filed
    2019 : Non-provisional 16277485 filed
         : Inventors assign to Apple Inc
    2020 : Pre-grant publication 2020 0082155
    2021 : Patent US 10990805 granted
    2039 : Listed adjusted expiration

Only two distinct ownership-relevant actors appear across the whole chain: the two inventors and Apple Inc. There is no second assignee to place on the timeline.

---

NPE / troll-pattern signals

For each of the eight signals I evaluated under the task rubric:

1. Shell-entity transfer — NOT PRESENT. No transfer to any licensing-only LLC. The chain is inventors → Apple Inc. and terminates there; the assignee of record is the original operating company. No "IP / Patents / Licensing / Holdings / Ventures" suffix appears in the entire chain.

2. Known asserter in the chain — NOT PRESENT. The only assignee ever recorded is Apple Inc. Apple does not appear on any of the listed NPE directories (Acacia, Marathon, IV, IPNav, Wi-LAN, Conversant/Mosaid, Vringo, Pendrell, Innovatio, MPHJ, Lumen View, Round Rock, Document Generation Corp, Spangenberg entities). Consistent with the earlier Litigation and PTAB sections of this analysis, which found no assertion and no AIA proceeding touching the '805 patent.

3. Repeat correspondent across the chain — NOT PRESENT (and not assessable). With a single recorded link, there is no "chain" over which recurrence could arise, so the signal cannot be present by definition. I could not retrieve the correspondent for even that one link (see gap note), so I record this as not present / not assessable — not as "clean."

4. Cascading transfers through chained LLCs in <24 months — NOT PRESENT. Only one assignment event (2019-02-15) exists; there are zero subsequent transfers, so no cascade is possible.

5. Pre-litigation transfer — NOT PRESENT. There is no litigation naming this patent (per the earlier Litigation and PTAB sections), so there is no suit for a transfer to precede. The sole assignment is contemporaneous with filing, not with any enforcement.

6. Bankruptcy fire-sale — NOT PRESENT. Apple is not and has not been in bankruptcy; no Chapter 7/11 proceeding involves this patent. No court-ordered transfer appears.

7. Privateering — NOT PRESENT. No transfer from Apple to any NPE asserting on Apple's behalf. No such transfer is recorded, and no NPE assertion of this patent was found.

8. Defensive aggregator — NOT PRESENT (and not the correct category). The chain does not terminate at RPX, AST, LOT, Unified Patents, or OIN. It terminates at the original operating company, which is a stronger non-assertion posture than defensive-aggregator neutralization — the patent was never transferred at all.

---

Verdict

Insufficient data. Putting aside the routine inventor→employer assignment, there are no post-issuance assignments, no reel/frame or correspondent verified for the one recorded link, and no litigation or AIA record for US 10,990,805. All eight NPE/troll signals are not present or not assessable, and — critically for this rubric — a "not present" call on signals 3 and 5 rests partly on records I could not retrieve rather than on verified absence. That evidentiary gap forces the "insufficient data" verdict rather than a confident "operating-company" or "non-asserting" call.

Two sentences of justification: The only conveyance of record is a 2019-02-15 inventors→Apple Inc. assignment (Kalscheur and Susskind), recorded the same day the non-provisional was filed, with no reel/frame or correspondent retrievable from the sources I could reach — so even the single link cannot be fully documented. There is no second assignee, no NPE-suffixed entity, no security interest, and no cascading or pre-litigation transfer, which is the classic profile of an operating company retaining its own patent; the verdict is "insufficient data" only because the reel/frame-and-correspondent gap prevents me from certifying the record as complete rather than because any negative or positive NPE signal was found.

Verify at (recommended):

• USPTO Assignment Center / Patent Assignment Search — search by patent number 10,990,805 or application 16/277,485: https://assignmentcenter.uspto.gov/ (legacy: https://assignment.uspto.gov/patent/index.html)
• Google Patents legal events for US10990805B2: https://patents.google.com/patent/US10990805/en#legalEvents

What would change the verdict: retrieving the reel/frame + correspondent for the 2019-02-15 record (to close signals 3 and 5), and a USPTO Patent Center transaction-history check confirming no later recordation exists. If those confirm a single inventor→Apple assignment and nothing else, the verdict should be upgraded to Defensive / non-asserting (chain terminates at the original operating company, never transferred).

I'll research the prior-art references cited on the face of US 10,990,805, using its Google Patents "Patent Citations" section and related sources.

Let me try to retrieve the specific cited-references list from the patent's front page and Justia's citation tables.

I found the examiner-cited reference list. Let me verify it against the granted patent's front page (56) and retrieve any foreign/non-patent citations, plus the "Cited By" table that shows which of these were applied.

Prior-Art Report — U.S. Patent No. 10,990,805

Scope, method, and hard limits (read first)

• Method: I have web search only — no direct PATENTSCOPE/PatentCenter/CourtListener API. I retrieved the examiner-cited ("Referenced Cited" / front-page (56)) U.S. patent-document list for US 10,990,805 from the Justia reproduction of the granted patent, and cross-checked individual items where search surfaced them. I did not retrieve the (56) foreign patent documents subsection or the non-patent literature subsection, and I could not confirm which of these references the examiner actually applied in a rejection versus merely listed after consideration.
• This list may be non-exhaustive. The snippet I retrieved showed 16 U.S. patent documents. Treat the list below as the best available enumeration, not a certified copy of the front page. The authoritative source is the front-page PDF at https://patentimages.storage.googleapis.com/0c/11/33/41ce2f726b0e8e/US10990805.pdf and USPTO PatentCenter for application 16/277,485.
• Anti-fabrication rule applied: I state the bibliographic data as retrieved. Where I could not verify a reference's subject matter in this session, I say so and do not invent a description. Those cells are marked [unverified].
• Effective filing date / §102 boundary: priority is US Provisional 62/730,115, filed 2018-09-12; non-provisional filed 2019-02-15. Under AIA §102(a)(1), a reference must be publicly available before 2018-09-12 to be anticipatory art; under §102(a)(2), a U.S. patent/application publication must be effectively filed before 2018-09-12. Two items below (US 10,303,866; US 10,402,094) issued in 2019, after the filing date, so they can only be §102(a)(2) art via an earlier effective filing — or were cited for other reasons (e.g., by the applicant/related-app prosecution).

Contradiction to flag

Your task header states the current date is April 26, 2026, while the earlier sections of this analysis and the system date say 2026-09-28. That discrepancy is unresolved and is carried forward from the prior sections. It does not affect the prior-art analysis below (all art predates 2019), but it does affect the litigation chronology discussed earlier.

---

1. The patent's printed citation list (U.S. patent documents)

Verbatim as retrieved from the Justia "Referenced Cited" table at https://patents.justia.com/patent/10990805#3:

---

2. §102 assessment, reference by reference

I map each reference to the claim elements I set out in the earlier sections (granted claim 1 = unlock request → simultaneous flood + patterned IR → ≥2 images → flood IR data → combine into composite → depth map from composite → authorize on both; granted claim 9 = device claim with illuminator(s) and a capture series in which at least one image is captured un-illuminated).

---

3. What the cited art does not appear to reach

Independent claim 9 — the device claim requiring a captured series in which at least some images are captured under both illuminations and at least one image is captured while the user is not illuminated (the "empty frame" of FIGS. 9–10, verified by the pseudo-random-sequence/empty-frame-verification process of FIG. 10) — has no obvious §102 counterpart in the retrieved citation list. The empty-frame liveness/replay-defense feature (and the "second device / overlapping illumination" verification of FIG. 10) is the clearest point of novelty versus the cited art. If anything bears on it, it is the spec's own cross-cited Prakash applications 16/049,933 and 16/050,021 — which are the applicant's own later-filed applications and therefore not §102 art against a 2018-09-12 priority.

Similarly, no single retrieved reference appears to disclose the full combination of claim 1 (simultaneous dual illumination + multi-image composite depth + dual-path feature-vector scoring against separate flood and depth template spaces per claims 6–7). Gren is the closest on illumination and dual-path comparison; Kilpatrick is the closest on compositing. A §103 combination of those two — same field, both concerned with deriving 2D texture and 3D geometry from one capture session — would be the obvious attack, and would be a stronger theory than either alone.

---

4. Art the specification itself identifies (not on the front-page list)

These are material to a full prior-art picture and should be included in any invalidity/FTO study, though they are "incorporated by reference" background rather than examiner citations:

• US 8,150,142 (Freedman et al.), US 8,749,796 (Pesach et al.), US 8,384,997 (Shpunt et al.), and US 2016/0178915 (Mor et al.) — cited in the '805 spec as examples of depth-map imaging from patterned illumination. These go to dependent claims 2–3 (pattern assessment → depth) and to claim 1's "generate depth map image data."
• US application Ser. No. 15/912,917 (Gernoth et al.) — cited for speckle/dot pattern density selection; relevant to claim 2's "areas between features in the patterned infrared illumination."
• US application Ser. Nos. 16/049,933 and 16/050,021 (Prakash et al.) — cited for anti-spoofing liveness security processes; relevant to claim 9's empty-frame architecture. Note: these are the applicant's own applications; check their filing dates against 2018-09-12 before treating them as art.

---

5. Confidence summary and recommended verification

To close the gaps, query, in order:

1. The front-page (56) list in the granted patent PDF and in USPTO PatentCenter for application 16/277,485 (this captures the foreign documents and NPL I could not retrieve, plus any items truncated from the Justia snippet).
2. The prosecution history (file wrapper) for 16/277,485 — this identifies which cited references the examiner applied and which the applicant distinguished, i.e., the real §102/§103 record. My list shows what was cited, not what was applied.
3. Google Patents #patentCitations for US10990805B2 to confirm the same list and to capture any European/JP/KR/CN documents.

Bottom line: I can confirm the examiner-cited U.S. patent-document list for US 10,990,805 (16 items) and can substantiate an anticipatory theory for one of them (Gren US 2016/0260223) against independent claim 1 and its dependents, with Kilpatrick US 7,961,332 supplying the composite-image element for a §103 combination. For the remaining entries, I have bibliographic data but not verified disclosures, and I have deliberately not supplied descriptions I cannot support. The granted claims' distinguishing features relative to this art appear to be (a) the express multi-image composite depth generation folded into claim 1 and (b) the un-illuminated frame of claim 9.

I'll ground the §103 analysis by verifying the subject matter of the key cited references before combining them. Let me run targeted searches.

Obviousness Analysis — U.S. Patent No. 10,990,805 under 35 U.S.C. § 103

Built on the Prior Art section above (16-item examiner-cited list) and the claim element map from the Patent Summary section. I re-ran targeted verification searches on the highest-value references and found one material contradiction with the prior section, flagged in §2 and §8. Where a reference's disclosure remains unverified, it is marked [unverified] and I do not build a ground on it.

---

1. Legal framework and working assumptions

• Statute: AIA §§ 102/103 apply (effective filing date post-2013-03-16). Priority is US Provisional 62/730,115, filed 2018-09-12; non-provisional US 16/277,485 filed 2019-02-15.
• Critical date: 2018-09-12 for §102(a)(1) (patented/published/public use before filing) and for §102(a)(2) ("effectively filed" before that date). If the provisional does not support the composite-image and empty-frame limitations that prosecution folded into the claims (see Patent Summary §"Independent claims"), the effective date slides to 2019-02-15 and more art becomes available — I address that contingency in §6.
• PHOSITA: a person with a B.S. in EE/CS/optics (or equivalent) and ~3–5 years' experience in IR/structured-light depth sensing and biometric imaging on mobile devices — the level implied by the citations themselves (Apple, Google, BASF, Microsoft/Nokia-origin IR-imaging inventors).
• Standard: KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007), and MPEP § 2143 rationales (A)–(G). A reference need not be in the cited list to be art; the cited list tells us what the examiner saw, not the universe of what exists.

---

2. Element map and reference capability matrix

Claim 1 (method): E1 unlock request on a UI; E2 illuminate with flood IR and patterned IR together; E3 capture ≥2 images under both illuminations; E4 generate flood IR image data from the ≥2 images; E5 combine the ≥2 images into a composite image; E6 generate depth map data from the composite image; E7 authorize via facial recognition on both flood data and depth data.

Claim 9 (device): D1 processor/memory/camera/UI + ≥1 illuminator providing both illuminations; D2 receive unlock request; D3 illuminate with both; D4 capture a series in which at least some images are under both illuminations AND at least one image is captured while the user is NOT illuminated; D5 generate flood + depth data; D6 authorize.

---

3. Grounds of rejection

Ground 1 — §102/§103 over US 10,204,262 B2 / WO 2018/132271 (the strongest single reference, and not on the '805 face)

This reference was not among the 16 citations the examiner listed, yet it maps to E1–E4, E6 (as structured-light-derived depth), E7, D1–D3, D5–D6 almost element-for-element: concurrent flood + structured IR illumination in one wavelength range, a combined image, extraction of the structured-light component, depth/3D comparison against 3D authentication data, flood/2D comparison against 2D authentication data, and a match decision for biometric anti-spoofing of a user.

• If the "at least two images … combine into a composite image" limitation (E5/E6) reads onto the combined-image + extracted-components disclosure, this is a §102(a)(1) anticipation via the WO 2018/132271 printed publication (published mid-2018, before 2018-09-12 — verify the exact date, which appears to be 2018-07-19) and §102(a)(2) via the US patent (effectively filed 2017-12-29).
• If E5/E6 are not met (because the reference derives depth from a component of a single combined frame rather than from a composite of ≥2 frames), the reference still renders claim 1 obvious when combined per Ground 2, because the only gap is the frame-fusion step — solved by Wan or FlashScan3D (§3.2, §3.3) with a predictable-result rationale.

Timing detail worth flagging: US 10,204,262 issued 2019-02-12, three days before the '805 application was filed (2019-02-15), and its WO counterpart published ~2018-07-19, months before Apple's provisional. So this art was publicly available before Apple's critical date but does not appear to have been before the examiner.

Ground 2 — §103: Gren US 2016/0260223 + Wan US 10,440,294 (and optionally + US 10,204,262)

1. Gren supplies E1–E4, E7 and D1–D3, D5–D6: simultaneous flood + structured illumination of a subject, capture of image(s), extraction of the flood-derived image and the structured-light-derived structure, comparison to 2D reference data and 3D reference data, and a match/no-match authentication — as evidenced by the ISA's "X" citation against claims 1–15 of the flood/structured-light family WO 2018/132271, whose disclosure I verified directly. Gren published 2016-09-08, squarely §102(a)(1) art.
2. Wan supplies the only missing element (E5/E6): "generating a composite image based on the first image and the second image" and "determining an estimated distance to an object … based on the composite image."
3. Motivation: both are in the same field (IR depth imaging on consumer devices); Wan's composite is expressly used for the same purpose the '805 claims (deriving depth from fused frames); result is predictable (averaging/stacking multiple frames reduces noise and fills depth holes — precisely the advantage the '805 specification itself recites at the FIG. 11 passage).
   • Fallback if Wan's 2018-09-05 filing date cannot be confirmed: substitute US 2007/0229850 A1 (FlashScan3D), which expressly discloses producing a composite image from texture + geometry captures and expressly names face recognition as an application (see Ground 3). Wan is a §102(a)(2) reference only (filed 7 days before Apple's provisional), so its effective date must be verified against a priority document if it is the primary composite reference.

Ground 3 — §103: Gren or US 10,204,262 + US 2007/0229850 A1 (FlashScan3D)

US 2007/0229850 (published 2007-10-04, §102(a)(1)) provides the composite-image teaching (E5) directly — "texture data … overlaid onto the 3D coordinates … to produce a composite image" — and, in the single-camera embodiment, provides the series-with-sparse-illumination teaching that is the closest thing in the cited art to claim 9's empty frame: the camera takes "a series of images" and the pattern flash fires with "only one or more images in the series." A POSITA combining FlashScan3D's compositing/projection architecture with Gren's or US 10,204,262's simultaneous flood + structured IR capture and dual 2D/3D biometric matching arrives at claim 1 without the exercise of invention; the combination is a mere substitution of one known illumination/capture architecture for another to obtain the predictable benefit of both data types from one session.

Ground 4 — §103 on the dependents

Ground 5 — §103 on claim 9's empty-frame limitation (D4) — weakest, and the likely non-obviousness redoubt

No reference located in the source list discloses capturing a series in which at least one frame is captured with the illumination(s) off and verifying a pseudo-randomly ordered image/empty-frame sequence. The best available building blocks are:

• US 2007/0229850 — a series of images in which the pattern flash is fired for "only one or more images in the series," i.e., some frames are captured without the structured illuminator. This is not the same as fully un-illuminated, but it establishes that deliberately capturing illumination-sparse frames within a multi-frame sequence was known.
• Motivation (MPEP 2143 (C) — known technique to improve a similar device): capturing frames with the source off is the classic dark-frame/background-baseline technique in imaging (dark-frame subtraction for sensor-noise and ambient characterization). A POSITA motivated to know what light is reaching the sensor independently of the device's own illuminators — precisely to detect a second device's overlapping illumination or a spoofing display — would predictably capture an un-illuminated frame.
• Motivation (anti-spoofing design incentive): anti-spoofing is the express purpose of both US 10,204,262 and Gren; a blank frame is a natural, predictable liveness probe (external illumination can be detected in which the device's own illumination is provably absent) and also guards against recombination/replay of stored illuminated frames.
• Randomization: the '805's use of a pseudo-random empty-frame placement is a routine variation of a known technique (challenge-response ordering) and, absent a reference teaching random ordering, is best attacked by combining with art on randomized capture/challenge sequences. I have not located such a reference in the source list.

Net: claim 9's D4 element (empty frame, randomized/verified) is the one feature for which I cannot construct a clean §102 ground and can construct only a moderately supported §103 ground. See §5.

---

4. Motivation to combine — consolidated rationales

1. Same field, same problem (MPEP 2143 (A)). Every reference above addresses IR/structured-light depth or 2D-IR biometric imaging. Gren, US 10,204,262 and the '805 all target anti-spoofing in face authentication; their stated problems are each other's.
2. Combination of known elements to obtain predictable results (MPEP 2143 (B)). Flood + pattern IR simultaneously yields two data streams from one session — predictable given Gren/US 10,204,262; fusing ≥2 frames into a composite to denoise and fill depth holes — predictable given Wan/FlashScan3D and expressly recited as the advantage in the '805 specification.
3. Known technique to improve a similar device (MPEP 2143 (C)). Multi-frame fusion with registration (the '805 names the iterative closest point algorithm — a textbook method, Besl & McKay 1992; the contemporaneous KinectFusion RGB-D fusion literature is of the same genre) and dark/blank-frame acquisition are both standard imaging techniques.
4. Explicit design incentive / market demand (MPEP 2143 (D), (F)). The '805's own Background concedes devices "separately capture[d] images using the different types of illumination to avoid one type of illumination affecting the other." That is a design constraint, not a teaching away: Gren and US 10,204,262 demonstrate the constraint is met by co-locating the sources in one IR band, balancing output, and extracting the structured component from the combined image. Once the constraint is shown solvable, the recognized benefit of simultaneous capture (no inter-frame motion misalignment, fewer exposures, single-sensor BOM) supplies the motivation.
5. "Obvious to try" (MPEP 2143 (G)). With the references identifying the problem and a finite number of identified, predictable illumination/capture configurations (alternating pairs; simultaneous single frame; simultaneous multi-frame), arriving at the claimed hybrid multi-frame scheme is a straightforward design-space optimization.

---

5. Where the patent is most likely to survive — and the counter-arguments to pre-empt

---

6. Confidence and the §102(b)(2)(C) wrinkle

(a) Confidence table

(b) The strongest overall theory is Ground 1/2 on claim 1 (US 10,204,262 or Gren alone, plus Wan/FlashScan3D for the composite step) and Ground 3/5 on claim 9 (with the empty frame the fragile element).

(c) §102(b)(2)(C) common-ownership screening. Several cited references are Apple's own (US 10,303,866 Van Os; and check US 9,612,687 Powell et al.). Any such reference that issued after the '805's filing is only §102(a)(2) art, and §102(b)(2)(C) excludes it if the subject matter and the claimed invention were commonly owned (or subject to assignment to the same person) by the effective filing date. This is likely to knock out part of the IDS as usable art and is a reason a petitioner should lead with non-Apple art (Gren, Wan, FlashScan3D, US 10,204,262) — which is also immune to any "same-inventor" §102(b)(1)(A) exception.

(d) Do not cite as art: US 2020/0218915 A1 (Samsung: flood + structure lighting devices with a randomly generated image-obtaining pattern) surfaced in my search but published in 2020, after the '805 critical date. It is not §102/§103 art. It is useful only as background showing that randomized capture-mode ordering was a recognized design space; using it as motivation would be impermissible hindsight.

---

7. Recommended verification steps before relying on this analysis

1. Pull the '805 granted claims and the 16/277,485 file wrapper (PatentCenter) to confirm the claim text and see which cited references were applied — including whether US 10,204,262 / WO 2018/132271 was ever before the examiner.
2. Retrieve the Gren (US 2016/0260223) specification and confirm the [0018], [0019], [0043] passages the ISA relied on, plus its assignee and effective filing date.
3. Confirm US 10,204,262's front page: application number, §371/continuation chain to PCT/US2017/068850, and exact grant date (2019-02-12) and WO 2018/132271 publication date.
4. Confirm US 10,440,294's effective filing date (application 16/122,224, 2018-09-05) and whether it claims priority to an even earlier document.
5. Verify the identities of the uncharacterized citations (Powell 9,612,687; Lau 2013/0015946; Coons 2013/0251215; Xu 2013/0343634; Zepeniuk 2015/0154035; Meir 2015/0341619; Satish 2017/0186160; Trail 2017/0195653; Price 2018/0196998; Ceylan 2018/0218535) — several are likely face-recognition feature-vector/scoring references usable against claims 6–7.
6. Search independently for art on blank/dark-frame capture and on challenge-response randomized capture ordering — the only path to a credible claim-9 rejection.

---

8. Contradictions and corrections to the earlier sections (flagged as required)

(a) US 7,961,332 — the prior section's characterization is contradicted. The Prior Art section stated, at "moderate confidence," that US 7,961,332 B2 (Kilpatrick et al., 2011-06-14) is the granted FlashScan3D structured-light/texture "composite image" patent. Live search results identify US 7,961,332 as "Fiber-optic heterodyne imaging vibrometer," James Kilpatrick, 2011-06-14, associated with Metrolaser — not a 3D-capture patent. I therefore do not use US 7,961,332 in any ground above, and substitute the verified US 2007/0229850 A1 disclosure (texture + geometry + pattern flash → composite image; face-recognition application) for the compositing teaching. Either Justia's entry or the identifier linkage is in error; resolve before relying on that citation.

(b) Gren vs. WO 2018/132271 — possible conflation in the prior section. The prior section attributed to Gren a flow of "Simultaneously illuminating flood light … → Collecting reflection image → Extracting reflection image generated from the flood light → Extracting … from the structured light … → Match or No Match," citing the WO 2018/132271 PDF. Those figures are the WO's own method, not necessarily Gren's; Gren appears in that document only as the ISA's category-X citation. The inference about Gren's disclosure is nonetheless sound in direction (an "X" citation against all of claims 1–15), but the textual attribution should be corrected: the verified disclosure is WO 2018/132271's; Gren's own text remains unread in this session.

(c) Date inconsistency (carried forward). This task header states the current date is 2026-04-26, while the earlier sections and system date use 2026-09-28. I have used 2026-09-28 consistent with the prior sections. The discrepancy does not affect §103 (all analysis is tied to 2018/2019 art dates) but does affect any litigation chronology.

(d) Prior-art direction between the '805 and the trinamiX/BASF family — needs care. The Litigation section reported that Apple's US 10,990,805 is cited as prior art by later trinamiX patents. The verified chronology above shows the opposite pressure for the flood + structured-light family: WO 2018/132271 / US 10,204,262 (PCT filed 2017-12-29, WO published ~mid-2018, US granted 2019-02-12) predates the '805's 2018-09-12 provisional by months. These may be different trinamiX families (the asserted patents in trinamiX v. Apple, No. 7:26-cv-00348, are 12,530,925; 12,288,421; 12,456,334; 12,361,760; 12,298,394; 12,332,352; 12,461,238 — none of which is US 10,204,262), so the two observations are not necessarily inconsistent, but the prior-art relationship between the two portfolios should be charted family-by-family rather than stated in one direction. Unresolved; verify.

---

9. Bottom line

• Claim 1 is vulnerable. A single reference — US 10,204,262 B2 / WO 2018/132271, which is not on the '805 face — discloses concurrent flood + patterned IR illumination, extraction of both flood-derived and structured-light-derived data from a combined capture, separate comparison to 2D and 3D authentication data, and an authentication decision. It is §102(a)(1) art via its WO publication (published before the 2018-09-12 priority date) and §102(a)(2) art via its 2017-12-29 PCT filing. The only claim-1 element not clearly met, compositing ≥2 images before deriving depth, is supplied by Wan (US 10,440,294) or FlashScan3D (US 2007/0229850) with a predictable-result / known-technique rationale, making a §103 combination strong even if the §102 reading of the composite limitation fails.
• The dependents largely fall with claim 1, with claims 2–4 and 8 tracking US 10,204,262's own extraction and dual-comparison disclosures, and claims 6–7 reduced to score-combination design choices.
• Claim 9 is the patent's real defensive value, and specifically its un-illuminated ("empty") frame within a randomized, verified capture series. No reference in the source list anticipates it; the §103 case requires importing dark-frame/blank-frame acquisition plus randomized capture ordering from art not yet located. Until such art is found, claim 9's empty-frame feature is where an invalidity challenge is most likely to stall — and, correspondingly, where an FTO analysis should concentrate.

Confidence in the overall verdict: moderate-to-high on claim 1 and the dependents; moderate on claim 9. The findings above rest on the retrieved disclosure of US 2007/0229850 and US 10,204,262 (claims + spec), the ISA citation record for Gren, and abstract-level data for Wan; the unverified entries in §2 are deliberately excluded from all grounds.