3D Reconstruction Patents: Top Companies & Filing Trends 2026
- Filings peaked in 2024 at 37, with the midpoint year 2022 at 23 — growth has flattened rather than kept accelerating.
- G06T covers 134 of 135 records, making core image-data processing near-universal while surgical, measurement and video-communication applications stay in single digits.
- Recent-year momentum is flat or negative across every tracked assignee, with no company showing renewed filing growth in the latest tracked year.
Filing growth compares 2021 (6 records) with 2024 (37) — a three-year span. 2024 is the most recent year we treat as complete: publication lags filing by roughly 18 months, so 2025 onwards are still filling in and any growth rate that ends there would understate the field. Top-5 share is the combined record count of the five largest assignees divided by all 135 records in scope (CR5), not by the ranked leaders only.
What this patent landscape covers
This landscape covers patent families filed under classifications spanning image data processing and generation (G06T), AI-based computing (G06N) and image/video recognition (G06V), where the claims specifically reference 3D reconstruction, neural radiance fields or gaussian splatting alongside technical concerns like multi-view consistency, rendering speed, sparse-view handling and geometry accuracy. It spans 135 published families filed between 2015 and the 2026 data cut-off, drawn primarily from United States, WIPO/PCT and European filings.
The dataset captures the shift from early photogrammetry-style reconstruction methods toward implicit neural scene representations and, more recently, explicit gaussian splatting — three distinct technical routes to the same end goal of producing a navigable 3D scene from 2D images.
Filing trends and technology composition
Filings in this dataset run from 2015 through the 2026 data cut-off, with publication lag meaning the most recent year is always undercounted relative to where filing activity actually stands.
A peak year followed by a flattening curve
Annual filings rose from 6 in 2017 to a peak of 37 in 2024. The midpoint year, 2022, sits at 23 records, which puts the growth trajectory closer to flat-to-declining than to a still-accelerating field once the partial, undercounted 2026 year is set aside.
Concentration in image-data processing
G06T (image data processing and generation) appears in all but one of the 135 records, making it close to a universal classification for this dataset rather than a differentiator. Secondary classification in G06N (AI-based computing) and G06V (image/video recognition) marks where reconstruction methods intersect with learned models and recognition pipelines, while H04N, G06F, G01B, G06K and A61B each register only single-digit counts, marking genuinely peripheral application areas.
Shares are the percentage of the 135 records in scope. A patent can carry several IPC classes, so the shares add up to more than 100%.
Go deeper on 3D Reconstruction and Neural Rendering with Eureka
This page is one run against one query. Ask Eureka your own question about 3d reconstruction and neural rendering and every answer comes back with the patent numbers behind it.
Try EurekaRepresentative filing and most-cited records
Point-based neural radiance field for three dimensional scene representation
A scene modeling system receives a plurality of input two-dimensional (2D) images corresponding to a plurality of views of an object and a request to display a three-dimensional (3D) scene that includes the object. The scene modeling system generates an output 2D image for a view of the 3D scene by applying a scene representation model to the input 2D images. The scene representation model includes a point cloud generation model configured to generate, based on the input 2D images, a neural point cloud representing the 3D scene, and a neural point volume rendering model configured to determine, for each pixel of the output image, the corresponding rendered value.Filed by Adobe Inc., published 2024-01-11 as US20240013477A1.
| # | Publication no. | Patent title | Citations |
|---|---|---|---|
| 1 | US20150178988A1 | Method and a system for generating a realistic 3D reconstruction model for an object or being | 222 |
| 2 | US20180315221A1 | Real-time camera position estimation with drift mitigation in incremental structure from motion | 52 |
| 3 | US20210279952A1 | Neural rendering for inverse graphics generation | 45 |
| 4 | US20240355047A1 | Three dimensional gaussian splatting initialization based on trained neural radiance field representations | 39 |
| 5 | US20210279943A1 | Systems and methods for end to end scene reconstruction from multiview images | 38 |
| 6 | WO2013174671A1 | A method and a system for generating a realistic 3D reconstruction model for an object or being | 37 |
| 7 | US20180315232A1 | Real-time incremental 3D reconstruction of sensor data | 36 |
| 8 | US20180315222A1 | Real-time image undistortion for incremental 3D reconstruction | 26 |
| 9 | US20190213789A1 | Use of temporal motion vectors for 3D reconstruction | 22 |
| 10 | WO2023080921A1 | Neural radiance field generative modeling of object classes from single two-dimensional views | 20 |
Citation counts are drawn from documents indexed within this search corpus and favour earlier filings; treat them as a measure of influence rather than current technical relevance.
Each row carries its publication number; clicking a row searches Eureka by that number.
Put your own technology through the same analysis
Eureka on the web
When you want the answer in the next five minutes.
The agent works the prompt against patents and technical literature, citing every source.
Run your analysis now →MCP server & REST API
When it has to run inside your own pipeline.
Patent search, landscape analysis and assignee resolution as MCP tools. Drop them into any agent framework, or call REST directly.
Browse MCP servers →What the filing pattern signals
Three figures from this dataset matter more than the raw record count for anyone deciding where to file or where to watch.
Growth has flattened since the midpoint
Annual filings rose from 6 in 2017 to 23 by 2022 and peaked at 37 in 2024. That is a slowing rate of increase, not a still-accelerating curve, even before accounting for publication lag in the final years.
Nearly every record sits in one IPC subclass
G06T (image data processing and generation) appears in all but one record, meaning classification alone does not differentiate filings in this space. G06N and G06V presence marks where AI-model and recognition claims overlap with reconstruction methods.
Influence skews toward older foundational filings
The most-cited record in the corpus is a realistic 3D reconstruction model, filed well before the current neural rendering wave. High citation counts here reflect age and foundational status more than current technical importance.
Eureka can read the same corpus for gaps instead of for coverage: under-claimed branches adjacent to 3d reconstruction and neural rendering, with the prior art for and against each one.
Who is filing, and where the field is open
Recent-year momentum across the tracked assignees is uniformly flat or negative, which is unusual for a technology area with an active 2024 filing peak — it suggests consolidation around existing claims rather than a new wave of entrants.
No tracked assignee is currently accelerating
Every named assignee with recent-year data shows flat or negative year-on-year filing change, from modest declines to a full drop to zero filings in the latest tracked year. That pattern is consistent with a maturing claim landscape rather than an emerging one.
Collaboration is limited and thin
Only eight co-assignee pairs appear across the corpus, each with a single shared filing. Co-filing is not a significant feature of how this field's IP is being built, which points toward in-house development over joint ventures or research partnerships.
Filing is concentrated in the US and PCT route
United States filings (73) and WIPO/PCT applications (37) together account for the large majority of records, with Europe, India, China and Hong Kong each contributing single-digit-to-mid-teens counts. That concentration matters for where enforcement risk and prior art density are highest.
| Assignee | Recent year | YoY |
|---|---|---|
| Google LLC | 1 | -50% |
| InterDigital VC Holdings, Inc. | 1 | -67% |
| NVIDIA Corporation | 0 | -100% |
| Magic Leap, Inc. | 0 | — |
| Samsung Electronics Co., Ltd. (Korea) | 0 | -100% |
| Intel Corporation | 0 | — |
| Lockheed Martin Corporation | 0 | — |
| Adobe Inc. | 0 | — |
Where to take this analysis
The filing and citation patterns above point to specific next steps depending on whether the goal is freedom-to-operate, competitive tracking, or identifying a filing opportunity.
Check freedom-to-operate against the anchor claims
The two most-cited records in this corpus, one on realistic 3D reconstruction and one on point-cloud-based neural rendering, define the broadest claim boundaries. Any commercial pipeline using explicit point clouds as an intermediate representation should be checked against these first.
Run a claim check in EurekaTrack assignee momentum for early signals of a new entrant
With every tracked assignee currently flat or declining, a shift to positive momentum from any single filer would be a meaningful early signal worth monitoring rather than a routine data point.
Set up momentum tracking in EurekaEvaluate the under-claimed application branches
Surgical, measurement and video-communication applications of 3D reconstruction show single-digit filing counts against a near-universal core classification, suggesting room for domain-specific claims that combine existing reconstruction methods with a narrower use case.
Explore white space in EurekaFrequently asked questions
Neural radiance field (NeRF) patents claim an implicit scene representation, typically a neural network queried per-ray to produce color and density values, which is accurate but historically slow to render. Gaussian splatting patents instead claim an explicit set of 3D gaussians rasterized directly, trading some representational flexibility for substantially faster rendering. In this dataset the newest splatting filings explicitly build on trained NeRF outputs for initialization, showing the two approaches are converging in claim language rather than remaining separate technical camps.
Filing activity in this dataset is spread across a mix of large technology companies and academic-adjacent assignees, with no single filer showing strong positive momentum in the most recent tracked year. Several of the more active historical assignees show flat or negative year-on-year filing counts, which is consistent with a field where core claim space is already occupied rather than one where a new leader is emerging. Reviewing the assignee ranking table alongside the momentum figures gives a more reliable read than any single company name.
Patent applications are typically published around 18 months after filing, so any record filed in the last year or so of a dataset's coverage window has not yet appeared in the public record. This dataset's cut-off is mid-2026, meaning the 2025 and 2026 figures understate real filing activity and should not be read as a genuine slowdown without accounting for that lag. The 2024 peak of 37 filings is a more reliable recent data point than the years immediately after it.
The technology composition data shows core image-processing classifications (G06T) present in nearly every record, while adjacent application areas — surgical and diagnostic imaging, dimensional measurement, and video-communication compression — each register only a handful of filings. That gap suggests the underlying reconstruction and rendering methods are well covered, but their application to specific domains like intraoperative imaging remains comparatively open. A first claim combining an existing scene representation method with a specific domain workflow is more likely to clear prior art than a broad reconstruction claim.
Not necessarily. A high density of filings means the claim space is occupied and that freedom-to-operate analysis is more important before building a product, but it does not by itself indicate the underlying technology has stopped improving or is commercially proven. In this dataset, filing activity peaked in 2024 and citation counts are skewed toward older, foundational records simply because they have had more time to accumulate citations, not because current work is less significant.
Research 3D Reconstruction and Neural Rendering in depth with Eureka
Go past this page: query the whole 3d reconstruction and neural rendering corpus yourself, in your own scope.
Every answer comes back with patent numbers you can open.
Disclaimer. This page is generated from Patsnap Eureka data drawn from a limited snapshot of global patent and scientific-literature records, and is provided for general information and reference only.
Patent data carries inherent limitations: recent filings (typically the most recent 18–24 months) are under-counted due to standard publication lag; counts may be reported at either a patent-family or a patent-record basis and are not always directly comparable; classification, applicant-name, and citation data may contain errors, duplicates, or omissions; and the underlying search query defines and constrains the scope shown. As a result, the analysis may be incomplete or inaccurate and may not reflect the full technology landscape.
Nothing on this page constitutes an exhaustive prior-art, novelty, freedom-to-operate, or validity search, nor does it constitute legal, financial, investment, or professional advice, and it should not be relied upon as such. Any patent, commercial, or strategic decision should be verified independently and reviewed with qualified patent, legal, and domain professionals. Patsnap makes no warranties, express or implied, as to the accuracy, completeness, or fitness for any particular purpose of the information presented.
Machine translation. Assignee and organisation names originally recorded in Chinese, Japanese or Korean have been rendered into English by an AI translation step so that the tables stay readable. These renderings are best-effort and may not match a company’s registered English name; the original name is what the underlying patent record carries, and it is what any Eureka query launched from this page uses.