US 9552633

I'll search for this specific patent number in patent databases and court dockets.

Let me search for the specific litigation history and any CAFC appeals involving this patent.

US Patent 9,552,633 B2 — Summary

Search scope note: I looked for this specific number (US 9,552,633 / US9552633B2) in patent databases and court/PTAB docket sources. I found no 2026 CAFC docket or appeal involving this patent. The 2026 Federal Circuit decisions that surfaced in my search involved other patents entirely (e.g., the Spero v. Mercedes-Benz IPRs, the Netlist appeals, GET v. Sony on U.S. 6,219,730). The Qualcomm–Apple Federal Circuit appeals that exist (e.g., Qualcomm Inc. v. Apple Inc., 24 F.4th 1367 (Fed. Cir. Feb. 1, 2022), Nos. 2020-1558/1559; and Apple Inc. v. Qualcomm Inc., 992 F.3d 1378 (Fed. Cir. Apr. 7, 2021)) concern different patents ('674, '037, '362), not the '633 patent. So: no evidence of a CAFC 2026 matter for 9,552,633.

Bibliographic data

Abstract (verbatim): "Systems, methods, and devices for enhancing an image are described herein. In some aspects, a device comprises a memory unit configured to store a left image and a right image. The left image and right image each depict a same scene from a different viewpoint. The device further comprises a coder configured to retrieve the left image and the right image from the memory unit. The coder is configured to determine a depth map based on a difference in spatial orientation between the left and right image. The device further comprises a processor coupled to the coder. The processor is configured to identify a portion of the left or right image selected by a user. The processor is further configured to determine an enhancement region surrounding the portion selected by the user based on the depth map. The processor is further configured to enhance the enhancement region."

Plain-language overview of the independent claims

There are four independent claims in the granted patent — claims 1 (method), 10 (apparatus, means-plus-function), 18 (non-transitory computer-readable medium), and 22 (device). All four share the same core five-step concept.

Claim 1 — Method
A method for enhancing an image that does the following:

1. Retrieve a stored left image and right image, where both depict the same scene from different viewpoints.
2. Compute a depth map from the difference in spatial orientation (disparity) between the left and right images.
3. Identify a user-selected portion of the left or right image, where that selected portion has a first depth value.
4. Determine an "enhancement region" around the user selection, based on the depth map — specifically, the region is the contiguous area extending from the selected portion that has a depth within a threshold of the selected portion's depth (i.e., neighboring pixels belonging to the same depth layer/object).
5. Enhance that enhancement region.

Claim 10 — Apparatus (means-plus-function)
The same concept expressed in "means for" language: means for retrieving the left/right images, means for determining the depth map, means for identifying the user-selected portion (having a first depth), means for determining the contour/threshold-bounded enhancement region, and means for enhancing that region. Because it is drafted in means-plus-function form, its scope is bounded by the corresponding structure disclosed in the specification (e.g., the encoder/motion estimation unit and the processor).

Claim 18 — Non-transitory computer-readable medium
The same five operations, but recited as code that, when executed, causes an apparatus to: retrieve the left/right images; determine the depth map; identify the user-selected portion (having a first depth); determine the contiguous, threshold-bounded enhancement region from the depth map; and enhance that region.

Claim 22 — Device
The device-level counterpart: a memory unit storing the left/right images; a coder that retrieves the images and determines the depth map from the difference in spatial orientation; and a processor coupled to the coder that identifies the user-selected portion (having a first depth), determines the contiguous threshold-bounded enhancement region from the depth map, and enhances that region.

Notable dependent-claim subject matter (from the 26 dependent claims): degrading/blurring the unselected portion (2–3, 11–12, 23–24); building the depth map via a disparity map that is upscaled (4, 13, 25); block-matching to find corresponding blocks within a threshold (5, 14, 26); growing the region until a depth discontinuity, a preset boundary, or a maximum distance is reached (6, 15, 27); and specific enhancement operations such as sharpening, moving/cutting the region, or adjusting color (7, 16, 28). Claims 8–9, 29–30 add wireless-device/cellular-transmission limitations (transmitting the depth map to a destination device).

Litigation / PTAB history (as reflected in the databases searched)

• Qualcomm Inc. v. Apple Inc., No. 3:17-cv-02398 (S.D. Cal., filed Nov. 29, 2017; Judge Dana M. Sabraw, Mag. J. Mitchell D. Dembin) — part of Qualcomm's Nov. 2017 five-patent assertion against Apple, where the '633 patent was tied to Apple's dual-camera "Portrait Mode" on iPhone 7 Plus / 8 Plus / X. (docketalarm)
• ITC Inv. No. 337-TA-1093, Certain Mobile Electronic Devices and Radio Frequency and Processing Components Thereof (II) — the '633 patent appears, with claim terms "coder" and "Order of method steps" identified as significant in the joint claim construction chart, and ALJ MaryJoan McNamara's Initial Determination issued March 26, 2019. (Joint Claim Construction Chart, May 15, 2018)
• IPR2019-00185 and IPR2019-00186 — Apple Inc. v. Qualcomm Incorporated (both filed Nov. 12, 2018, patent 9,552,633). Both were terminated by settlement, not decided on the merits: the Board's termination decision ("Dismissal of Petitions and Termination of the Proceedings," 37 C.F.R. §§ 42.71(a), 42.74) is dated May 1, 2019, with refunds approved Nov. 30, 2020. The Google Patents family page likewise flags these as "filed (Settlement)."
• Global context: Qualcomm and Apple settled worldwide litigation in April 2019, which is consistent with the IPR terminations.

Uncertainty / caveats

1. No CAFC 2026 docket found. My search surfaced no Federal Circuit appeal in 2026 (or any year) on this patent. If a 2026 appeal exists, it was not indexed in the sources I could reach; I cannot confirm one.
2. Claim-text version. The pre-grant publication (US 2015/0254811) and the granted patent differ in claim numbering/structure (e.g., the published version's independent claims are 1, 8, 16, 24; the granted version's are 1, 10, 18, 22). My overview above follows the granted claim set.
3. "Assignee" vs. real-party status. The databases list QUALCOMM Incorporated as assignee throughout; the ITC/court records confirm Qualcomm as complainant/plaintiff. The patent's annual-fee/legal status is listed as Active with the adjusted expiration date noted above — treat that as a database assumption, not a legal conclusion.
4. Everything in the bibliographic block above (title, inventors, dates, abstract, claims) is corroborated by multiple independent sources (Google Patents, Justia, FreePatentsOnline, Docket Alarm, and the patent PDF itself).

Sources: Google Patents US9552633B2; Justia patent 9552633; FreePatentsOnline 9552633; Docket Alarm patent search.

I'll research the litigation history for US Patent 9552633. Let me search multiple sources.

Let me search for more specific details on the ITC investigation and the district court case.

Let me verify the IPR details and the final disposition of the district court case.

Litigation Involving US Patent No. 9,552,633

Patent identified: US 9,552,633 B2 — "Depth aware enhancement for stereo video" (Assignee: Qualcomm Incorporated; inventors Shilpi Sahu and Mainak Biswas; priority/filing date March 7, 2014; granted January 24, 2017).

I searched litigation databases (Unified Patents portal data, Docket Alarm, PACER/CourtListener, ITC records) for this specific number. The '633 patent was asserted by Qualcomm against Apple only, in the 2017–2019 Qualcomm–Apple global patent dispute. I found no other litigation involving this patent. (Note: I interpreted the number literally as 9,552,633 and disregarded look‑alike numbers such as '356, '336, '674, '002, which are related but distinct Qualcomm patents asserted in the same campaigns.)

The known matters are:

1. District court — Qualcomm Incorporated v. Apple Incorporated

• Plaintiff: Qualcomm Incorporated
• Defendant: Apple Incorporated
• Jurisdiction: U.S. District Court for the Southern District of California
• Case No.: 3:17-cv-02398-DMS-MDD (Judge Dana M. Sabraw; Magistrate Judge Mitchell D. Dembin)
• Filing date: November 29, 2017
• Role of the '633 patent: The complaint asserted five Qualcomm patents, including the '633 patent, directed at, among other things, "depth-data-based image enhancement." Qualcomm alleged Apple's dual-camera iPhones (iPhone 7 Plus, 8 Plus, and X, via "Portrait Mode") infringed the '633 patent. Qualcomm sought injunctive relief, damages, and fees.
• Outcome/status: Resolved by the global settlement announced April 16, 2019, in which Qualcomm and Apple agreed to drop all litigation worldwide, Apple made a payment to Qualcomm, and the parties signed a six-year patent license (effective April 1, 2019) and a multiyear chipset supply agreement. The case was dismissed. (Case filings remained active into late 2018/early 2019 — e.g., Apple's answer and counterclaims filed January 22, 2018, and claim construction briefing in 2018 — before the settlement.)

2. ITC Section 337 investigation — Certain Mobile Electronic Devices and Radio Frequency and Processing Components Thereof (II)

• Complainant: Qualcomm Incorporated
• Respondent: Apple Inc. (Commission Investigative Staff also a party)
• Jurisdiction: U.S. International Trade Commission
• Investigation No.: 337-TA-1093
• Filing/Institution dates: Qualcomm's complaint was filed November 2017 (investigation instituted January 18, 2018; 83 Fed. Reg. 834, Jan. 8, 2018)
• Presiding ALJ: MaryJoan McNamara
• Role of the '633 patent: The complaint asserted the '633 patent (originally claims 1–3, 10–12, 18, and 22–24), along with the '356, '336, '674, and '002 patents. The '633 patent claims were withdrawn during the investigation; by the final initial determination the investigation had narrowed to the '356, '336, and '674 patents only.
• Outcome/status: Final Initial Determination (March 26, 2019) found a violation of Section 337 only as to claim 1 of the '674 patent (not the '633 patent). Following the same April 2019 Qualcomm–Apple settlement, the Commission granted a joint motion and terminated the investigation in its entirety on the basis of settlement (notice signed August 20, 2019).

3. PTAB inter partes reviews — Apple Inc. v. Qualcomm Incorporated

• Petitioner: Apple Inc.
• Patent Owner: Qualcomm Incorporated
• Jurisdiction: USPTO Patent Trial and Appeal Board
• Case Nos.: IPR2019-00185 and IPR2019-00186
• Filing date: November 12, 2018
• Role of the '633 patent: Both petitions were filed by Apple against Qualcomm in November 2018 and are indexed in litigation databases under US 9,552,633; at minimum, IPR2019-00186 is confirmed as directed to the '633 patent. (I could not conclusively confirm whether IPR2019-00185 also targets the '633 patent specifically versus a related Qualcomm patent; the database search for patent number 9552633 returned both dockets, so I am reporting both rather than dropping one.)
• Outcome/status: Both proceedings were terminated before institution. The Board issued a "Decision — Dismissal of Petitions and Termination of the Proceedings" (37 C.F.R. §§ 42.71(a), 42.74) dated May 1, 2019, and issued "Notice of Refund" documents on November 30, 2020. This outcome is consistent with the parties' global settlement (the Google Patents family record likewise lists both IPRs as filed and subsequently settled). Because the petitions were never instituted, there was no PTAB final written decision and no Federal Circuit appeal.

Summary

• Total distinct litigation matters tied to US 9,552,633: three — one district court case (S.D. Cal. 3:17-cv-02398), one ITC Section 337 investigation (337-TA-1093), and two PTAB IPRs (IPR2019-00185 and IPR2019-00186).
• All were against Apple Inc. No third-party defendants, and no CAFC appeal, were found for this patent.
• All were resolved by the April 16, 2019 Qualcomm–Apple global settlement (dismissals/terminations; no merits judgment against Apple on the '633 patent — the only ITC violation finding in that campaign was on the different '674 patent).

Caveats on confidence: The district court/institutional facts (case number, judge, filing date, settlement) and the ITC investigation number and disposition are well documented (USITC notice; CourtListener/PACER docket; Qualcomm/Apple press releases) and I am highly confident in them. The specific attribution of IPR2019-00185 to the '633 patent (vs. a related Qualcomm patent) is the one point I could not fully verify from the sources retrieved; treat that attribution as provisional. If you need absolute precision, I recommend pulling the PTAB docket for IPR2019-00185 directly from the USPTO Patent Trial and Appeal Board (or Unified Patents' PTAB portal) to confirm the challenged patent number.

AIA Trial Proceedings — U.S. Patent No. 9,552,633 B2

Patent: US 9,552,633 B2 — Depth aware enhancement for stereo video (Sahu et al.; Qualcomm Incorporated). Filed 2014-03-07; granted 2017-01-24; 30 claims; adjusted expiration listed as 2035-05-19. Source: Google Patents US9552633B2.

⚠️ Data conflict flagged up front. The structured "PTAB proceedings on file" block states the USPTO ODP API returns no AIA trial proceedings. That is contradicted by the patent's own Google Patents litigation record and by Docket Alarm, both of which show two IPR petitions on the '633 patent (IPR2019-00185, IPR2019-00186). I am reporting the two proceedings I could independently corroborate, and flagging the ODP gap. I did not find any other AIA proceeding (no PGR, no CBM, no second wave).

Proceedings overview

Two AIA proceedings on file, both Inter Partes Review, both by [Apple Inc.](/litigations/by-plaintiff/Apple%20Inc.) against Qualcomm Incorporated, and both terminated pre-institution on 2019-05-01 by joint dismissal in view of the global Apple–Qualcomm settlement — zero instituted, zero Final Written Decisions, zero claims canceled, zero claims sustained.

Breakdown: active: 0 · claims invalidated: 0 · claims sustained: 0 · settled/terminated pre-institution: 2 · institution denied: 0.

Bottom line for a defendant: this patent has never been substantively tested at the PTAB. All 30 claims remain live, unamended, and unadjudicated. The good news for a defendant is the corollary: no § 315(e) estoppel exists (estoppel attaches only on a Final Written Decision), and Apple's petition record — including the prior art it assembled — is public and reusable. The defensive posture is not "the patent is hardened"; it is "the patent is untested, and the road map is on the shelf."

IPR2019-00185 — Apple Inc. v. Qualcomm Incorporated

• Type: Inter Partes Review
• Filed: 2018-11-12 (petition filed; Docket Alarm docket date). Note: Qualcomm's infringement complaint against Apple, S.D. Cal. No. 3:17-cv-02398, was filed 2017-11-29 (Judge Dana M. Sabraw), which places Apple's filing roughly 11.5 months later — consistent with a § 315(b) one-year-deadline-driven filing.
• Status: Terminated — "Dismissal of Petitions and Termination of the Proceedings," 37 C.F.R. §§ 42.71(a), 42.74 (PTAB, 2019-05-01, Decision Document No. 9). Google Patents tags the case "(Settlement)". Plain English: the parties jointly asked the Board to dismiss, the Board did, and nothing was ever decided on the merits.
• Judge panel: Not publicly confirmed. No institution decision issued, so there is no panel-authored decision naming APJs. A Docket Alarm query that returned this docket carried a filter for APJ Trevor Jefferson; treat that as a weak, unverified signal, not a confirmation of the panel.
• Petition grounds: Not verified from the public record I could retrieve. The petition and its exhibits are on PTAB E2E, but I could not pull the claim list or the § 102/§ 103 theory from an authoritative source. I will not guess. Do not cite a ground for -00185 without pulling the petition itself.
• Institution decision: None. The proceeding terminated ~5.5 months after filing, before the six-month § 314(b) institution deadline. No § 314(a) analysis was ever published.
• Final Written Decision: None issued. Consequently no independent claim, no dependent claim, and no claim of the '633 patent was canceled or upheld. The patent stands exactly as granted.
• Settlement / termination: Joint dismissal resulting from the worldwide Apple–Qualcomm settlement announced 2019-04-16 — all litigation between the companies dismissed, a six-year patent license effective 2019-04-01 (with a two-year extension option), a multi-year chipset supply agreement, and a one-time payment from Apple to Qualcomm (amount undisclosed). A Notice of Refund issued 2020-11-30 (Document No. 11), returning the unearned post-institution fees. Terms of the PTAB dismissal itself are confidential; the paper reflects the global deal. See Apple press release, 2019-04-16.
• Appeal: None. A pre-institution termination is not a final, appealable agency action to the Federal Circuit, and no CAFC appeal exists.
• Defensive value: Neutral-to-good. Because the case died pre-institution, there is no estoppel and no adverse claim construction binding anyone. A defendant today can lift Apple's prior-art selection wholesale and re-file without facing § 315(e) or the Board's earlier § 325(d) work-product.
• Docket: Docket Alarm — IPR2019-00185; Unified Patents PTAB record.

IPR2019-00186 — Apple Inc. v. Qualcomm Incorporated

• Type: Inter Partes Review
• Filed: 2018-11-12 (same filing date as -00185; the two were filed as parallel petitions).
• Status: Terminated — settlement / dismissal. Google Patents tags the case "(Settlement)". Same disposition as -00185.
• Judge panel: Not publicly confirmed — no institution decision, therefore no panel-authored paper.
• Petition grounds: Partially corroborated only. The only ground evidence I could verify is an exhibit designation: Exhibit 1005-3 = U.S. Patent Application Publication No. 2013/0208093 A1 (Sun et al., Aptina Imaging Corp., "System for reducing depth of field with digital image processing," published 2013-08-15) — a § 103-style obviousness reference aimed a year before the '633 priority date. The claim-by-claim mapping and the full ground set are not verified; pull the petition from PTAB E2E before relying on it.
• Institution decision: None. Terminated pre-institution.
• Final Written Decision: None issued. No claim of the '633 patent was canceled or confirmed.
• Settlement / termination: Same global Apple–Qualcomm settlement (announced 2019-04-16); joint dismissal.
• Appeal: None.
• Defensive value: Same as -00185 — a free, public prior-art lead (Sun '093) with no estoppel baggage and no merits ruling to overcome.
• Docket: Docket Alarm — IPR2019-00186; Apple's Exhibit 1005-3 is indexed at Docket Alarm exhibit page.

Strategic summary

Claim status. There is nothing to report on the merits, and that is the headline. No claim of US 9,552,633 has been canceled, amended, disclaimed, or held unpatentable in any AIA trial. All 30 claims remain as granted — 30 live / 0 canceled / 30 untested. If you have previously seen a representation that "claims 1–5 were canceled" or that the patent was "narrowed through IPR," that is wrong for this patent; the two proceedings died before the Board ever reached the merits, so no certificate under § 318(b) issued. Likewise, if someone asserts the patent "survived two IPRs and is therefore hardened," that is also wrong — survival by settlement is not survival on the merits, and a patent can be settled out precisely because the petitioner's art was credible.

Estoppel landscape. § 315(e)(2) estoppel keys off a final written decision under § 318(a). Neither IPR produced one, so no statutory estoppel applies to Apple or its privies, and none applies to anyone else. Practical consequences for a defendant being asserted against today: (i) every prior-art ground, including anything Apple used in 2019, remains fully available to you and to Apple; (ii) there is no § 325(d) legacy — the Board never issued a § 314(a) discretionary-denial analysis that a later petitioner must distinguish; (iii) the only bars you must clear are your own § 315(b) one-year clock and § 315(a)(1) if you filed a declaratory-judgment action first. In short: the ground floor is wide open.

Pattern signals. (1) Same petitioner, multiple IPRs: yes — Apple filed two parallel petitions on the same patent on the same day, the classic two-track strategy for maximizing institution odds (or a seven-figure word-count workaround). Both went away together. (2) Patent owner aggressiveness at the PTAB: Qualcomm did not appeal anything here because there was nothing to appeal; the cases ended by Qualcomm's own business settlement, not by PTAB outcome. (3) Defensive aggregator: no. Google Patents renders "Petitioner: 'Unified Patents PTAB Data'" next to these dockets — that is a CC-BY data-license attribution on the litigation data, not a stakeholder. Unified Patents was not the petitioner; Apple Inc. was. Do not let the metadata mislead an estoppel or privity analysis. (4) Enforcement posture: the '633 was asserted by Qualcomm (as patent owner) against Apple's dual-camera "Portrait Mode" on iPhone 7 Plus / 8 Plus / X, in S.D. Cal. 3:17-cv-02398 and ITC Inv. No. 337-TA-1093, alongside four unrelated Qualcomm patents (7,693,002; 8,063,674; 9,154,356; 9,473,336). This is a patent-owner enforcement patent, not a troll patent — which matters because Qualcomm has the resources to litigate it and just demonstrated it is willing to assert it.

Recommended next steps

1. Pull the two petitions from PTAB E2E before doing anything else. The most valuable artifacts in this file are Apple's claim-by-claim grounds and exhibits. I could confirm only one reference (Sun '093, Apple Ex. 1005-3 in -00186). The full ground sets are not reproduced in any secondary source I could verify. Start at USPTO PTAB E2E / PTAB public search and search IPR2019-00185 and IPR2019-00186 by patent number 9,552,633.
2. Read the termination papers verbatim rather than relying on the "(Settlement)" tag. The operative document is IPR2019-00185, Paper 9, dated 2019-05-01 — "Dismissal of Petitions and Termination of the Proceedings, 37 C.F.R. §§ 42.71(a), 42.74." The § 42.74 citation means the parties filed a settlement agreement and requested termination; confirm whether a copy of the agreement was filed (it is often confidential and not made of record), and whether the dismissals were with or without prejudice — that distinction affects re-filing strategy.
3. If you are a defendant facing assertion of the '633: do not brief or argue that any claim has been invalidated — you would be sanction-baited for misrepresenting the record. Your correct framing is: the patent is wholly untested at the PTAB; Apple's 2019 petitions were settled away pre-institution; here is our own § 102/§ 103 case. There is no FWD to quote and no disposition to link, precisely because none exists.
4. Watch your § 315(b) clock. If you have been served with a complaint alleging infringement of the '633, the one-year bar runs from service of the complaint, not its filing. Because the patent is untested, a well-built IPR is unusually attractive here — but only if filed inside the window, and only if the art is materially better than what Apple assembled in 2018.
5. Verify the ODP gap. The structured block's "no PTAB activity" answer is demonstrably incorrect for this patent (two settled IPRs from 2018–2019, docketed and terminated in 2019). If you are relying on ODP-derived reports, treat pre-2020 terminated AIA cases as a known blind spot and cross-check against PTAB E2E and CourtListener before stating that a well-asserted patent has "no PTAB history."

Confidence and limits. High confidence on: existence, petitioner, patent owner, filing date, termination date, termination basis, and the absence of any FWD or appeal. Low confidence / expressly not verified: the APJ panel identity, the challenged claim numbers, and the full ground sets in either petition. I have not asserted any of those.

I'll research the assignment record for US 9552633 before writing anything.

Let me verify the specific assignment record and litigation details.

Let me try to pin down the specific assignment reel/frame and confirm no post-issuance transfers.

Assignment & Ownership Analysis — US 9552633 B2

Patent: Depth aware enhancement for stereo video · App. 14/201,261 · Filed 2014‑03‑07 · Granted 2017‑01‑24
Source of record for ownership: Google Patents legal events (mirrors USPTO Assignment Center) — https://patents.google.com/patent/US9552633/en
Verification entry point: https://assignmentcenter.uspto.gov/ (search by patent number 9552633)

Data limitation, stated up front: I could confirm only one recorded assignment event for this patent (the inventor‑to‑corporate assignment of 2014‑06‑19). The publicly cached legal‑events record does not expose the reel/frame number for that entry, and I was unable to retrieve it from the sources available. I have not invented a reel/frame. Where a field is unknown, it is marked not retrievable rather than guessed. Anyone needing the reel/frame for a filing must pull the Assignment Center abstract directly.

Inventors

Pattern assessment: No unusual pattern. Both are Qualcomm employees who assigned to their employer as a condition of employment — the normal case for a corporate‑filed application. There is no evidence of inventors departing near filing, and no inventor‑held continuation assignments, which is what a pre‑fire‑sale posture would typically show. The application was filed 2014‑03‑07 and the assignment recorded roughly three months later, consistent with standard onboarding/employment‑agreement practice rather than a distress transaction.

Original assignee

Qualcomm Incorporated — Delaware corporation, 5775 Morehouse Drive, San Diego, CA 92121‑1714. Qualcomm is the assignee on the face of the issued patent and remains the current assignee per Google Patents.

• Primary line of business: Fabless semiconductor design and licensing — Snapdragon mobile platforms, modems/RF front‑end, and a very large patent licensing program (Qualcomm stated >130,000 issued patents and applications worldwide in its 2018 litigation filings).
• Product embodiment: Qualcomm ships silicon (Snapdragon SoCs and Spectra/camera ISPs) that perform multi‑frame/depth‑aware image processing of the general type claimed. The '633 disclosure itself is framed around re‑using an existing on‑device encoder/decoder to compute depth, i.e., a hardware/software feature that ships in devices — it is not a purely abstract licensing asset.
• Status: Operating, publicly traded (NASDAQ: QCOM). Not in bankruptcy, not acquired, not dissolved.

Assignment timeline

Chronological, from the only recorded entry I could verify:

• 2014‑03‑07 (filing date) — no assignment recorded at filing; Qualcomm was the applicant of record.
• executed ~2014‑06 (exact execution date not retrievable) / recorded 2014‑06‑19 — Reel/frame not retrievable from cached sources; verify via Assignment Center
  • Conveyance: Assignment (Assignment of Assignors' Interest)
  • Assignor: Shilpi Sahu; Mainak Biswas (joint)
  • Assignee: QUALCOMM INCORPORATED, 5775 Morehouse Drive, San Diego, CA 92121
  • Correspondent: Not retrievable from the cached record. Note: Qualcomm's patents commonly list correspondence at "Qualcomm Incorporated — International IP Administration, 5775 Morehouse Drive, San Diego, CA 92121‑1714," but that is the patent correspondence address, not the recorded assignment correspondent, and I will not conflate the two. No repeat‑correspondent finding can be made without the abstract.
  • Context: Internal — inventor‑to‑employer assignment (the standard employment paperwork), not a transfer to a third party.
• 2017‑01‑24 — Patent granted; no post‑issuance assignment recorded.
• No further recorded assignments. Google Patents legal events show nothing after 2014 other than foreign‑family priority filings (KR, CN, EP, JP, BR) and the grant itself. Those are prosecution events, not ownership changes.

Bottom line: Aside from the single inventor‑to‑corporate assignment, the Assignment Center effectively has no post‑issuance assignment record for this patent — which is itself the finding: ownership never left Qualcomm.

Off‑record (non‑ownership) events bearing on the chain

These are not assignments and must not be mistaken for them, but they explain why this patent appears in NPE‑adjacent datasets:

• 2017‑11‑29 — Qualcomm sues Apple, Qualcomm Incorporated v. Apple Incorporated, No. 3:17‑cv‑02398‑DMS‑MDD (S.D. Cal.), asserting the '633 patent alongside U.S. 9,154,356; 9,473,336; 8,063,674; and 7,693,002. Case terminated 2019‑04‑23 (global settlement).
• 2017‑11‑30 — Qualcomm files an ITC complaint asserting five patents incl. '633 → Investigation 337‑TA‑1093.
• 2018‑11‑12 — Apple petitions for IPR on the '633 patent: IPR2019‑00185 and IPR2019‑00186 (apple as petitioner, Qualcomm as patent owner). Both settled.
  • ⚠️ The Google Patents page shows a "Petitioner: 'Unified Patents PTAB Data' by Unified Patents" string. That is the data‑provider license attribution for the PTAB dataset, not the petitioner. Docket Alarm's docket confirms the petitioner of record was Apple Inc. Do not read the Unified Patents string as Unified Patents being a party or an assignee here.

Timeline diagram

timeline
    title Ownership of US 9552633
    2014 : Filed by Qualcomm
         : Inventors assign to Qualcomm
    2017 : Patent granted
         : Qualcomm sues Apple in S D Cal
         : ITC case 337-TA-1093 filed
    2018 : Apple files IPRs 185 and 186
    2019 : IPRs settle
         : District court case terminates

NPE / troll-pattern signals

Verdict

Operating‑company assertion.

Justification: the ownership record contains a single assignment — the 2014‑06‑19 inventor‑to‑corporate transfer of Sahu and Biswas's rights to Qualcomm Incorporated — with no subsequent recorded assignment, so the patent has been held continuously by an operating, publicly traded semiconductor company rather than passed to any licensing vehicle. Qualcomm then asserted the '633 patent in its own name against an actual competitor (Apple) in Qualcomm v. Apple, No. 3:17‑cv‑02398 (S.D. Cal., filed 2017‑11‑29) and in ITC Investigation 337‑TA‑1093, and the resulting Apple IPRs (IPR2019‑00185/186, filed 2018‑11‑12) settled without any ownership change. None of the eight NPE signals is present, and the absence of a post‑issuance assignment record is affirmative evidence that the patent never left its original owner.

Verification link: https://assignmentcenter.uspto.gov/ — search patent number 9552633; confirm reel/frame for the 2014‑06‑19 "Assignment of Assignors' Interest" and confirm that no later conveyance is on file.

I'll research the prior art citations for US Patent 9,552,633. Let me search multiple sources.

I found the front-page reference list. Let me gather detailed information on each cited reference to accurately describe them.

Prior Art of Record for US Patent 9,552,633 B2 — "Depth aware enhancement for stereo video"

Important framing correction before the lists: The task asks which references "potentially anticipate under 35 U.S.C. § 102." Having reviewed the actual prosecution record (the file history submitted as Apple's Exhibit 1002 in IPR2019‑00186), no reference was ever applied as a § 102 anticipation rejection against any claim of the '633 patent. Every substantive rejection during prosecution was a § 103 obviousness rejection built from combinations of references (Niu + Shimoyama, further in view of Roderick). I flag this so the analysis is not overstated: individually, none of the cited references anticipates a claim; each is relevant only as an element of an obviousness combination. I therefore identify, for each reference, the claim(s) it was actually used against (in § 103) rather than fabricating a § 102 mapping.

A. The three references the Examiner actually applied

These are the most relevant prior art of record. They were cited by the Primary Examiner (Gregory M. Desire, Art Unit 2668) and form the only rejections on the merits. All quoted reasoning below is from the July 11, 2016 Non‑Final Office Action (Application/Control No. 14/201,261).

A1. Niu — US 2012/0008852 A1

• Full citation: U.S. Patent Application Publication 2012/0008852 A1, "System and Method of Enhancing Depth of a 3D Image," Sheng‑Chun Niu et al., assignee Himax Media Solutions, Inc.
• Publication / filing date: Published Jan. 12, 2012; filed May 18, 2011 (Appl. No. 13/110852).
• Description: Receives a left image and a right image of a 3D/stereoscopic scene; a depth generator produces one or more depth maps; a depth enhancer improves the depth map(s) by depth histogram stretching (applying a gain to depth levels) and/or by segmenting the depth map into sub‑maps and applying local depth histograms. It then enhances the original left/right images. Niu's own ancestor application (US 2012/0293488) confirms it is a Himax stereo‑image line of work.
• Claim(s) it was applied against: The Examiner asserted Niu taught the core of independent claims 1, 8, 16, and 24 (retrieving left/right images; generating a depth map; identifying a "portion... selected by a user"; determining an enhancement region; enhancing) and dependent claims 7, 14, 23, 30 (sharpening), 15 (encoder/processor structure for the means‑plus‑function claim), and 21 (selected portion "comprises a first depth"). The Examiner expressly conceded Niu "does not clearly disclose determining depth map based on difference in spatial orientation" — which is why the Shimoyama reference was pulled in.
• § 102? No anticipation. Niu lacks the spatial‑orientation/disparity derivation and the threshold‑bounded contiguous region. § 103 only, in combination.

A2. Shimoyama — US 9,053,575 B2

• Full citation: U.S. Patent No. 9,053,575 B2, "Image Processing Apparatus for Generating an Image for Three‑Dimensional Display," Kenichi Shimoyama et al., assignee Kabushiki Kaisha Toshiba. (Pre‑grant pub. US 2012/0069009 A1.)
• Publication / filing date: Granted June 9, 2015; application 13/234,448 filed Sept. 16, 2011; pre‑grant publication Mar. 22, 2012.
• Description: Detects an object in a 2‑D input image; selects a depth template corresponding to the object type and places/modifies it (based on the object's size and orientation) on a depth map; corrects the depth map by weighted smoothing (bilateral/ε/median filtering); and generates parallax images from the corrected depth map and input image. It also generates disparity maps from disparity templates.
• Claim(s) it was applied against: Combined with Niu against 1, 7‑8, 14‑16, 21, 23‑24, 30. The Examiner relied on Shimoyama specifically for "depth map based on difference in orientation" (citing the depth‑template‑modifying unit 500 and the depth‑map generating unit 200), to supply the element Niu lacked.
• § 102? No anticipation. Shimoyama does not disclose the user‑selected contiguous threshold region. § 103 only.

A3. Roderick — US 2010/0164952 A1

• Full citation: U.S. Patent Application Publication 2010/0164952 A1, "Stereoscopic Image Production Method and System," Michael Roderick (assignee/small entity; later associated with Woodman Labs/GoPro lineage).
• Publication / filing date: Published July 1, 2010; filed Mar. 7, 2010 (Appl. No. 12/718978); priority provisional 61/297,816 filed Jan. 25, 2010.
• Description: Produces left/right‑eye images from 2‑D footage. Objects are identified (rotoscoping, chroma/luminance keying, tracking), assigned relative depths in a gray‑scale layer to build a depth map, a composite is blurred (blurring tool) and the depth map is used by a displacement/distortion tool to displace the image left and right. Relevant to the "degrade/blur the unselected portion" limitations.
• Claim(s) it was applied against: Combined with Niu + Shimoyama against dependent claims 2‑3, 9‑10, 17‑18, 25‑26 — specifically the "degrading" (claims 2, 9, 17, 25) and "blurring" (claims 3, 10, 18, 26) limitations.
• § 102? No — used purely as a secondary § 103 reference.

Examiner‑cited U.S. references listed on the PTO‑892 (Notice of References Cited), by examiner, beyond the above:

None of these four was applied in a rejection; they appear on the Notice of References Cited as considered references.

B. Applicant‑cited U.S. references (IDS of record on the front page)

These appear in the "References Cited" block but were not relied on by the Examiner against the claims:

I was not able to retrieve full specifications/abstracts for Nowak, Sandrew, Chen, Sun (2013/0033582), and Hubyalkar in this session; I flag that their descriptions above are limited to class/title‑level information, and I have not verified deeper content.

C. Applicant‑cited foreign patent documents

D. Applicant‑cited non‑patent literature (IDS of record)

These are the depth‑map‑upsampling / stereo‑matching references relied on conceptually for the dependent‑claim features the Examiner found allowable (claims 4‑6, 11‑13, 19‑20, 22, 27‑29):

E. The claims the Examiner could NOT invalidate (the "allowable" nucleus)

For completeness — and because it defines what the prior art of record does not reach — the Examiner stated (Paper dated July 11, 2016, ¶¶ 6‑7) that no prior art was found for:

• Claims 4, 11, 19, 27: determining a disparity map based on the difference in spatial orientation (larger difference = higher disparity); upscaling the disparity map; and determining the depth map from the upscaled disparity map.
• Claims 6, 13, 22, 29: identifying a continuous portion from the selected point having depth within a threshold of the first depth, and growing the region until a depth discontinuity, preset boundary, or threshold distance is reached.

These became the amendments that carried the independent claims (1, 8, 16, 24 at the time; now 1, 10, 18, 22 in the granted patent) to allowance.

F. The most sophisticated invalidity theory of record (never tested on the merits)

In IPR2019‑00186 (Apple Inc. v. Qualcomm Inc.), Apple assembled the strongest obviousness challenge to the '633 patent. Those petitions were dismissed by settlement before institution (Joint Motion to Dismiss, Apr. 30, 2019; dismissal under 37 C.F.R. §§ 42.71(a)/42.74), so no ground was ever adjudicated. Apple's asserted references are therefore not "prior art of record" against the patent in the § 103 sense — no merits decision exists. The key exhibits were:

• Niu (US 2012/0008852) and Shimoyama (US 9,053,575) — the same primary references the Examiner used (Exhs. 1009, 1010).
• Sun (US 2013/0208093) and Zhang (US 2011/0096832) (Exhs. 1005, 1008).
• Koo & Jeong, "Fast Stereo Matching Using Block Similarity," ICCSA 2004 (Exh. 1007) — block‑matching, relevant to claims 5/14/26.
• Lynch (US 2013/0101175), Bhaskaran (US 2013/0251281), Chen (US 2013/0038686) (Exhs. 1006, 1015, 1026).
• Region‑growing/segmentation and selective‑enhancement NPL: Adams et al., Seeded Region Growing (1994); François et al., Depth‑Based Segmentation (1997); Chang et al. (1997); Fjørøft et al. (1996); Meier et al. (1996); Du et al. (2010); van der Schaar et al. (2001); Pourazad et al. (2010); Koschan et al. (1997) — all aimed at the "continuous, threshold‑bounded enhancement region" and "region‑based selective enhancement" limitations.
• Standards/treatise background: Haskell (MPEG‑2), Sonka et al., Castleman, Gonzalez & Woods.

Treat this section as advocacy‑stage prior art, not established invalidating art.

G. Caveats and confidence

1. § 102 vs. § 103 (repeat): The premise that these references "anticipate" claims is not supported by the record. Every rejection was a § 103 combination. If the goal is a § 102 chart, no cited reference alone anticipates any claim of the '633 patent.
2. Two "Sun" and two "Chen" families appear — the front page lists US 2013/0033582 (Sun) and US 2012/0293624 (Chen), while Apple's IPR used US 2013/0208093 (Sun) and US 2013/0038686 (Chen). These are different documents; I have kept the identifiers literal and not merged them.
3. Descriptions for the applicant‑cited set (Section B) are partly title/class‑level only; I could not retrieve full text for Nowak, Sandrew, Chen, Sun‑2013/0033582, and Hubyalkar this session. Treat those descriptions as provisional.
4. The "Note of References Cited" list (Section A table) reflects what the Examiner formally considered; the fuller "References Cited" front‑page block (Sections B‑D) is the IDS of record. Do not conflate the two.
5. No contradiction found with the previously generated summary/litigation sections — the IPR dismissal facts here (IPR2019‑00185/‑00186, Apple v. Qualcomm, settled) are consistent with those sections.

Primary sources: US 9,552,633 B2 front page (Google Patents / patentimages PDF); File History of the '633 Patent (Apple Exh. 1002, submitted in IPR2019‑00186, via Docket Alarm), including the July 11, 2016 Non‑Final Office Action and PTO‑892; Apple IPR2019‑00186 Exhibit List and Joint Motion to Dismiss (Apr. 30, 2019); US 2012/0008852 A1; US 9,053,575 B2; US 2010/0164952 A1.

I'll research the specific prior-art references cited on this patent's Google Patents page so the § 103 analysis is grounded in the actual record.

I have the actual "References Cited" list from the '633 patent's face. Let me pin down the key ones for the combination analysis.

Obviousness Analysis — U.S. Patent No. 9,552,633 B2 ("Depth Aware Enhancement for Stereo Video")

Builds on the earlier "Patent summary" and "Litigation summary" sections. That content is not repeated except where I need to reconcile it.

Reconciliation note (no contradiction, one update): The earlier litigation section flagged the attribution of IPR2019-00185 to the '633 patent as "provisional." The record now confirms it: the IPR2019-00185 docket is captioned [Apple Inc.](/litigations/by-plaintiff/Apple%20Inc.) v. Qualcomm Incorporated with exhibit "US Patent No. 9,053,575 to Shimoyama" (Ex. 1010, IPR2019-00185), and the Patent Owner Preliminary Response states "Case IPR2019-00185 / U.S. Patent No. 9,552,633" (POPR, IPR2019-00185). So both IPRs target the '633 patent (two Apple petitions against one Qualcomm patent). No merits ruling issued in either — both were dismissed on settlement, so the § 103 questions below were never adjudicated on the merits by the Board.

1. Legal framework and governing date

• Statute: The application was filed March 7, 2014 — after the AIA's first-inventor-to-file date (March 16, 2013) — so AIA 35 U.S.C. § 103 governs, with prior art defined by § 102(a)(1) (publicly available before the effective filing date) and § 102(a)(2) (U.S. patents/applications "effectively filed" before that date).
• Standard: Graham v. John Deere, 383 U.S. 1 (1966), as refined by KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007). The inquiry is whether the claimed subject matter as a whole would have been obvious to a PHOSITA, considering (a) scope/content of prior art, (b) differences, (c) skill level, and (d) objective indicia. In litigation the challenger bears the burden of clear and convincing evidence.
• Analogous art: References are combinable if from the same field of endeavor or reasonably pertinent to the problem. All the cited art here sits in stereo/multi-view image processing, depth-based image enhancement, and 2D→3D conversion — the same field.
• A material prosecution fact: The examiner's actual § 103 ground on the record is below; the applicant appears to have amended the independent claims to incorporate the "continuous portion within a threshold" limitation previously found only in dependents.

2. Person having ordinary skill in the art (PHOSITA)

A bachelor's degree in EE, CS, or imaging science plus ~2–5 years in image/video processing, stereo vision, or video coding — or a master's with less experience. The '633 specification itself frames the problem in codec terms (reusing H.264/HEVC motion estimation to compute disparity), so a PHOSITA here would be comfortable with both block-matching stereo and standard video-encoder architecture. That matters for the "coder" limitation of claim 22.

3. The limitations that carry the inventive weight

The five-step concept is common to claims 1, 10, 18, and 22. Only three elements are arguably non-routine:

1. Compute a depth map from the L/R spatial-orientation difference (disparity) — ubiquitous by 2014.
2. Identify a user-selected portion, where the selected portion has a first depth, and determine the region as "a continuous portion … starting from the selected portion … that has a depth within a threshold of the first depth" (granted claims 1, 10, 18, 22; see KR counterpart text at KR20160130414A).
3. "Coder" — flagged as a most-significant term in the ITC Joint Claim Construction Chart, Inv. 337-TA-1093 — along with "Order of method steps." If "coder" is construed (as the specification's FIG. 3 encoder / motion-estimation unit suggests) to require depth computed using video-coding motion vectors, that narrows the combinations available to a challenger. This is the claim term most likely to resist a broad § 103 attack.

4. The prior-art record actually on this patent

A. References cited on the face of the '633 patent (patent PDF, field (56); Justia):

* Published after the 3/7/2014 filing date; usable only as § 102(a)(2) art to the extent their effective filing dates precede 3/7/2014.

B. The examiner's actual § 103 ground (from the '633 file history, Ex. 1002, IPR2019-00186):

• "Claims 1, 7–8, 14–16, 21, 23–24 and 30 are rejected under 35 U.S.C. 103 as being unpatentable over Niu and Shimoyama… It would have been obvious … to substitute Niu['s] depth map generator with Shimoyama's depth map generator. The suggestion/motivation … would yield higher quality 3D image."
• "Claims 2–3, 9–10, 17–18 and 25–26 are rejected under 35 U.S.C. 103 as being unpatentable over Niu and Shimoyama … and further in view of Roderick (2010/0164952)… Roderick discloses degrading [a] portion of the left image or right image (note fig. 1, block 112)."

Two things are significant. First, the rejected claim numbers (independents 1, 8, 16, 24) correspond to the pre-grant publication claim set — corroborating the earlier section's caveat that the published and granted claim sets differ. Second, the examiner's combination covered the independent claims plus the "enhance/blur" dependents, but the excerpt I retrieved begins at rejection "4," so earlier items (likely covering the disparity-map, upscaling, block-matching, and continuous-portion/threshold dependents) are not in this excerpt. The granted independent claims now recite the "continuous portion … within a threshold of the first depth" language that in the published set appeared only in dependents 6/13/21/29 — consistent with an amendment made to overcome the Niu+Shimoyama rejection (labeled as inference, not verified against the full file wrapper).

5. Combinations that would render the claims obvious

Ground 1 — Niu + Shimoyama + Roderick (the ground of record)

Element mapping: Niu supplies a depth-map–driven enhancement of a 3D image (the "enhancing the enhancement region" step). Shimoyama supplies an image-processing apparatus that detects an object, generates a depth map, corrects depth values against image edges, and generates output images — supplying "determin[ing] a depth map" and the depth-selective notion of treating objects differently. Roderick supplies degrading/blurring the non-selected portion (dependent claims 2–3, 11–12, 23–24).

Motivation: Both Niu and Shimoyama are in the same field (3D/parallax image generation and depth-based image improvement); the examiner's stated rationale — "would yield higher quality 3D image" — is a predictable improvement rationale of the kind KSR endorses. A PHOSITA seeking a better-performing depth-based enhancement pipeline would naturally substitute the better depth-map generator. Roderick's blur/degrade teaching is a known photographic technique (background defocus) routinely applied in depth-aware processing.

Weakness: Shimoyama derives its depth map from depth templates placed on detected objects in a single 2D image, not from L/R spatial-orientation difference. The examiner characterized Shimoyama as teaching "depth map based on difference in orientation" — that reading is contestable. Likewise, Niu/Shimoyama do not clearly teach a user-selected point or a contiguous threshold-bounded region grown from that point. This is very likely the gap the applicant closed by amendment, and it is where a validity defense would concentrate.

Ground 2 — EP 2634772 (object selection) + a stereo-depth reference + a depth-refinement reference + a depth-based-enhancement reference

• EP 2634772 A1 (BlackBerry, 9/2013) — "Methods and devices for selecting objects in images": supplies the user-selected portion and the object/region delineation step.
• Mei et al. (2011) and/or US 9,053,575 / US 2012/0008852 / US 2011/0122230 (Boisson) — supply disparity/depth from stereo and block matching, covering claims 4, 13, 25 ("determin[ing] a disparity map … a larger difference in spatial orientation … indicates a higher disparity" and "determining a distance between a block in the left image and a corresponding block in the right image … identical within a threshold value").
• Kim et al. (2011; 2013) and Lee et al. (2012) — supply upscaling/refining the disparity map into a depth map (claims 4, 13, 25).
• Hachicha et al. (EUSIPCO 2012) and Li et al. (3DSA 2010) — expressly teach combining depth information with local edge detection for stereo image enhancement — the very purpose of the '633 patent — covering "enhancing the enhancement region" and supplying the edge-boundary rationale for stopping region growth at a depth discontinuity (claim 6).

Motivation: All four strands address the same problem (better-looking images from cheap stereo capture) with a finite, predictable set of known techniques; the '633 specification itself admits the framework: "The idea is to select an object in a picture using a cursor. An algorithm separates out the object or the region from the rest of the image based on the depth map and applies different degrees of enhancement or different filters based on the depth" (patent PDF, ¶0090). An applicant admission of this kind is powerful evidence that the combination was within ordinary skill. The specification also concedes that selection and depth-based enhancement were known, and that the alleged advantage was merely doing it more cheaply by reusing hardware ([¶0024], [¶0030] — "the existing system can be reused for computing depth"). Cost/implementation-efficiency advantages are classic KSR "predictable results" territory.

Ground 3 — Niu + Shimoyama + Roderick + Hachicha + Kim (2013)

Adds explicit depth-map upscaling and depth-weighted, edge-aware filtering to Ground 1, closing the gaps for claims 4–6, 13–15, 25–27. Hachicha's "local edge detection" supplies exactly the "until a portion … is identified that does not have a depth within a threshold" stopping criterion (depth discontinuity = object boundary), and Kim's upsampling supplies "upscaling the disparity map."

Ground 4 — US 2014/0368506 A1 (Abbott & Morse) as a near-anticipatory/secondary reference

Abbott & Morse, "Depth-Aware Stereo Image Editing Method, Apparatus and Computer-Readable Medium" (US 2014/0368506 A1; granted 10,109,076; provisional 61/834,017 filed June 12, 2013) (Justia) is the closest single reference located, and it is conspicuously absent from the '633 face citations. It discloses: receiving a stereo image (left + right); extracting depth/disparity from it; user application of depth-dependent effects (highlighting, relighting, hazing, blurring); and, critically — "the user selects a particular point or region on an image to select a reference depth for blurring operations. For example, pixels which have similar depths may remain in focus while pixels with dissimilar depths may be blurred proportional to the difference in depth with the reference depth."

That is, in substance, claim 1 plus claims 2–3 of the '633: a user-selected reference point → depth used to separate in-focus vs. blurred content → depth-proportional enhancement. Because its provisional predates the '633's March 7, 2014 filing (and the non-provisional was filed June 12, 2014, after), it qualifies as § 102(a)(2) prior art as of June 12, 2013 to the extent the provisional supports the relied-upon disclosure. Combined with Kim/Lee (upsampling) and Mei (block-matching disparity), it renders the independent claims and most dependents obvious; on its own it is strong § 103 art and a plausible § 102(a)(2) anticipation theory for claims 1–3 if the provisional support holds.

Ground 5 — "Coder" limitation (claims 22, 10, 17)

The '633 expressly reuses the video encoder's motion-estimation path to obtain disparity (patent PDF ¶¶0062–0063, 0082–0087). Wiegand et al.'s H.264 survey (cited on the face) supplies the encoder architecture, and the specification's own FIGS. 3–5 supply the motion-vector-as-disparity technique. Where a reference (e.g., any MVC/stereo-video codec disclosure) teaches coding an interleaved L/R stream, computing disparity from inter-coded motion vectors, that element is met. This is the element most vulnerable to a § 112(f) construction fight, but for § 103 it is well-supported by the codec prior art plus the applicant's own admissions.

6. Claim-by-claim summary

7. Where a validity defense would push back

1. Shimoyama's depth source. The examiner's reading of Shimoyama as teaching orientation-based (stereo) depth is not airtight; a defense can argue Niu+Shimoyama gives no stereo disparity and no user selection.
2. The "continuous portion within a threshold of the first depth" limitation. If the priority claim to stereoscopic depth is preserved, this region-growing criterion is the most specific limitation and is not squarely disclosed by Niu/Shimoyama. But Abbott & Morse, Hachicha, and EP 2634772 collectively undercut it.
3. "Coder." A § 112(f) construction tying "coder" to the disclosed encoder/motion-estimation structure would force a challenger to find that structure in the art — narrower but still meetable via Wiegand + stereo/MVC codec art.
4. Order of steps (flagged by the ITC). If construed as a required order, the challenger must show the references practice that sequence. Most do.
5. No teaching away. There is no evidence the art taught away from combining depth-based segmentation with user-directed selection; the art in fact converged on it (see "Cited By" of the '633, e.g., JP2023008998A (2021, Viewworks), which generates high-depth-of-field images from a stereo image — the same problem a decade later, at Google Patents family page).

8. Objective indicia (secondary considerations)

The record shows Qualcomm advanced copying and commercial-success/battery-benefit arguments, but they were disputed both at the ITC and in the S.D. Cal. case: Qualcomm's own damages experts' reports are cross-referenced on "value and importance" of the accused technology (Qualcomm trial brief, 3:17-cv-02398), and Apple argued the benefits testimony was duplicative of the ITC record (Joint motion memo, Dkt. 208-1). No tribunal ever ruled on the merits of validity for the '633 patent — the IPRs settled before institution (May 1, 2019 termination) and the litigation settled April 16, 2019. That means the objective-indicia record is incomplete and a settlement-based dismissal carries no preclusive effect on § 103.

9. Bottom line

• Independent claims 1, 10, 18, 22 are, in my assessment, more likely than not obvious — most compellingly over US 2014/0368506 A1 (Abbott & Morse, § 102(a)(2) as of 6/12/2013), alone or combined with the cited stereo-matching/upsampling art (Mei, Kim, Lee); alternatively over the Niu + Shimoyama + Roderick ground of record as supplemented by Hachicha and EP 2634772. The '633 specification's own admissions deprived the applicant of the "recognition of the problem" contribution that might otherwise support patentability.
• Dependent claims add little patentable weight: each element (upscaling, block-matching, edge-stopping, sharpen/move/recolor, wireless transmission) is disclosed or suggested by the face citations.
• Confidence is moderate, not high, because (a) I could not obtain the complete file wrapper and so am inferring the amendment strategy, and (b) "coder" and "order of steps" remain unconstrued and could narrow the claims in a way that improves their validity position. The strongest residual defense is a narrow § 112(f)-limited reading of "coder" plus the region-growing/threshold limitation, but Abbott & Morse and Hachicha materially weaken even that.

Caveats: Reference dates and disclosures are taken from the patent's face, the Google Patents family pages, and the IPR exhibit set; the file-history excerpt used for Ground 1 is partial (rejections 4–5 only), and the amendment inference is clearly labeled as such. No CAFC or merits ruling on this patent exists, so nothing here is a legal conclusion — it is a technical § 103 assessment.