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 →Filing growth compares 2021 (160 records) with 2024 (395) — 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 3,028 records in scope (CR5), not by the ranked leaders only.
This landscape covers patent families addressing document understanding, layout analysis and table structure recognition, combined with claims on reading order, chart data extraction, text region detection, coordinate grounding, multipage document handling or optical character recognition. The scope spans 3,028 published records filed or published between 2015 and the 2026-07-31 data cut-off, drawn from receiving offices led by the United States, followed by Europe, WIPO, China, India and Australia.
The technology sits at the intersection of classical OCR pipelines and newer AI-driven layout models, which is visible in the IPC mix: general digital data processing and image recognition classes dominate, while narrower application classes such as healthcare informatics remain thin. That pattern points to a field still being claimed largely at the infrastructure level rather than the vertical-application level.
Two views of the same 3,028-record dataset: filings over time, and the IPC subclasses those filings carry.
Recorded filings moved from 26 in 2017 to a peak of 395 in 2024, with the 2021-to-2024 span alone showing a +147% increase. 2025 and 2026 figures will continue to rise as publications catch up with filing dates, given the roughly 18-month lag between the two; they should not be read as a slowdown.
G06F (electric digital data processing) appears on 50.6% of the 3,028 records and G06V (image/video recognition) on 40.4%. G06K (data recognition and presentation) and G06N (AI-based computing) follow at 29.7% and 19.5%. Because records can carry multiple IPC classes, these shares sum to well over 100% and should be read individually, each against the full 3,028-record base.
Shares are the percentage of the 3,028 records in scope. A patent can carry several IPC classes, so the shares add up to more than 100%.
This page is one run against one query. Ask Eureka your own question about document and chart understanding and every answer comes back with the patent numbers behind it.
Try EurekaPhysical page layout analysis for optical character recognition is performed. A physical page layout analysis method finds constituent parts of an image and gives an initial data-type label, such as text or non-text. Within the text data, connected components are identified and analyzed. Tab-stops are detected from groups of edge-aligned connected components. The detected tab-stops are used to deduce the column layout of the page by finding column partitions. The column layout is then applied to find the polygonal boundaries of and a reading order of regions containing flowing text, headings, and pull-outs.Filed by Google LLC, published 2012-08-21. Its claims sit squarely on tab-stop-based column detection and reading-order derivation — a technique still foundational to layout analysis pipelines.


| # | Publication no. | Patent title | Citations |
|---|---|---|---|
| 1 | US20100070448A1 | System and method for knowledge retrieval, management, delivery and presentation | 1,798 |
| 2 | US20120191716A1 | System and method for knowledge retrieval, management, delivery and presentation | 579 |
| 3 | US5999664A | System for searching a corpus of document images by user specified document layout components | 494 |
| 4 | US20160171336A1 | Realogram Scene Analysis of Images: Multiples for Scene Analysis | 364 |
| 5 | US5963966A | Automated capture of technical documents for electronic review and distribution | 322 |
| 6 | US5848184A | Document page analyzer and method | 313 |
| 7 | US20030202697A1 | Segmented layered image system | 305 |
| 8 | US20040146199A1 | Reformatting documents using document analysis information | 269 |
| 9 | US20110197121A1 | Effective system and method for visual document comparison using localized two-dimensional visual fingerprints | 257 |
| 10 | US20040145593A1 | Resolution sensitive layout of document regions | 232 |
Citation counts are drawn from documents inside this searched corpus and favour older filings; treat them as a measure of influence on later filers, not of current commercial relevance.
Each row carries its publication number; clicking a row searches Eureka by that number.
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 →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 →Three signals from the dataset that matter more for strategy than the raw counts alone.
The ranked leader holds 193 records, and the top 5 combined reach 648 records — 21.4% of all records in scope. That leaves the large majority of filing activity spread across a long tail of assignees, so a strong position at the top does not close off the field the way it might in a more consolidated technology.
Filings rose from 160 in 2021 to 395 in 2024. The 2025-2026 figures in the raw trend look lower, but that reflects publication lag of roughly 18 months rather than a real drop in filing activity — decisions should be based on the 2021-2024 trajectory, not the tail years.
General digital-processing and image-recognition classes dominate the claim set, while healthcare informatics (G16H) sits at only 4.1% of records. That gap suggests domain-specific document and chart understanding — clinical records, regulatory filings, industry-specific tables — is comparatively open ground.
Eureka can read the same corpus for gaps instead of for coverage: under-claimed branches adjacent to document and chart understanding, with the prior art for and against each one.
The ranking spans 100 companies counted by records in scope; recent-year momentum shows most of the largest historical filers pulling back rather than accelerating.
The top-ranked assignee holds 193 records against 83 for the fifth-placed filer and 71 for tenth place — a visible gap at the very top, but one that narrows quickly further down the ranking.
Multiple large historical assignees, including major software and enterprise-technology filers, show zero recorded filings in the latest year, with year-over-year declines of -91% to -100% among those still filing. This is consistent with publication lag rather than necessarily a strategic exit, but it does mean the visible leaderboard understates who is active today.
The dataset records only 10 co-assignee pairs, with the strongest links all involving the same corporate assignee paired with different named inventors. Cross-company joint filing is not a notable feature of this landscape yet.
| Assignee | Recent year | YoY |
|---|---|---|
| UiPath Inc | 1 | -98% |
| THE STRATEGIC COACH | 1 | -91% |
| Microsoft Corporation | 0 | — |
| Xerox Corp | 0 | — |
| Intuit Inc | 0 | -100% |
| Google LLC | 0 | -100% |
| Microsoft Technology Licensing LLC | 0 | -100% |
| Canon KK | 0 | -100% |
The dataset points to specific next steps depending on whether the goal is freedom-to-operate, portfolio positioning, or scouting.
With 21.4% of all records held by five assignees, a targeted claim-chart review against just those five records is a fast way to surface the highest-probability blocking art before deeper searching.
Run a claim comparison in Patsnap EurekaSeveral of the largest historical filers show sharp year-over-year declines. Confirming whether that reflects publication lag or an actual pivot away from the space matters for competitive monitoring.
Set up assignee monitoring in Patsnap EurekaChart-to-data extraction and multipage table stitching show comparatively thin coverage relative to the core OCR and layout classes. Drafting around a specific technical mechanism there is more likely to clear prior art.
Explore white space in Patsnap EurekaThe ranked leader in this dataset holds 193 records, well ahead of the fifth-placed assignee at 83 and tenth place at 71. That said, the top 5 assignees combined account for only 21.4% of the 3,028 records in scope, so no single company controls the field. A freedom-to-operate review should look past the leaderboard to the long tail of single- and double-digit filers, since a specific blocking claim is just as likely to sit there.
Recorded filings grew from 160 in 2021 to 395 in 2024, a +147% increase, and 2024 is the highest year on record. Figures for 2025 and 2026 appear lower in the raw trend, but that is expected: publication typically lags filing by around 18 months, so the most recent one to two years are always undercounted at the time of any snapshot. The safe reading is that the field kept accelerating at least through 2024, not that it is cooling.
By IPC subclass, general digital data processing (G06F) appears on 50.6% of the 3,028 records and image and video recognition (G06V) on 40.4%, making these the two dominant claim areas. Data recognition and presentation (G06K) and AI-based computing (G06N) follow at 29.7% and 19.5% respectively. Narrower application areas, such as healthcare informatics at 4.1%, are comparatively underdeveloped, which points to where domain-specific filings could still find open claim space.
US8249356B1, assigned to Google LLC and published in 2012, claims a method for physical page layout analysis that detects tab-stops from edge-aligned connected components to deduce column layout, then uses that layout to derive reading order across text, headings and pull-outs. It is a foundational technique for OCR-based layout analysis, specifically tab-stop-driven column detection. New filings that rely on materially different layout-detection mechanisms, such as learned visual-grounding models rather than tab-stop geometry, are less likely to be blocked by it directly, but any claim built around geometric column inference from connected components should be checked against it.
The clearest gaps sit in vertical applications rather than core techniques: healthcare informatics carries only 4.1% of records despite general document-processing classes covering half the dataset, and co-assignee collaboration is limited to just 10 recorded pairs, suggesting little joint development activity. Specific technical branches such as chart-to-data extraction from scanned figures, cross-page table stitching in multipage documents, and coordinate grounding for handwritten annotations also show comparatively thin claim density relative to the core OCR and layout classes. These are reasonable areas to scope a first claim rather than filing directly against the dense G06F and G06V territory.
Go past this page: query the whole document and chart understanding 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.
Help us improve
Something wrong, out of date, or missing on this page? Tell us and it goes straight to the team that maintains it.
This page