Perceptual Image Quality Assessment Patents: Who Leads 2026
- Filing has flattened, not grown. Filings peaked at 8 in 2019 and sit at 5 by the 2022 midpoint — this is a maturing claim space, not a fast-growing one.
- No assignee is currently active. Every tracked assignee, including the most-cited filers, shows 0 filings in the latest year — a sign the field has gone quiet or that recent filings simply haven't published yet.
- Claim density concentrates in G06T. 73 of 83 records sit in image data processing (G06T), far ahead of recognition (G06V, G06K) and healthcare imaging (G16H), which stay in single digits.
Filing growth compares 2021 (6 records) with 2024 (6) — 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 83 records in scope (CR5), not by the ranked leaders only.
What this landscape covers
This review covers patent families published between 2015 and 2026 that combine image quality assessment language — full-reference, no-reference and perceptual similarity metrics such as LPIPS and structural similarity — with classification in image processing (G06T7), image/video recognition (G06V10) and pictorial communication (H04N17). The 83 records captured span convolutional-network-based full-reference scoring, two-stage no-reference assessment, and distortion-type feature-distance approaches.
Filing is split across a small set of receiving offices, with the United States and China together accounting for most of the volume and a thinner tail through India, the WIPO PCT route, Europe and Australia. Publication lags filing by roughly 18 months, so the apparent drop in the most recent year understates real filing activity — but the multi-year plateau before that drop is a genuine signal.
Let an AI agent run this analysis on your own technology
Pick a task. Every answer cites the patents behind it.
Filing trend and technology composition
Two views of the same 83-record dataset: how filing activity has moved year over year, and how the underlying technology splits across IPC subclasses.
A plateau, not a growth curve
Filings rose to a peak of 8 in 2019, then settled near the 2022 midpoint of 5. Read against a 2026 count of 2 in a partial year, the honest interpretation is flat-to-declining activity rather than an emerging technology still accelerating.
Image processing dominates the claim space
G06T (image data processing & generation) appears in 73 of 83 records, well ahead of G06K, G06V and H04N, each in the low twenties. AI-model computing (G06N) and adjacent commercial or healthcare applications (G06Q, G08B, G16H) are present but thin, marking them as extension areas rather than the core of current filing.
Shares are the percentage of the 83 records in scope. A patent can carry several IPC classes, so the shares add up to more than 100%.
Go deeper on Perceptual Image Quality Assessment with Eureka
This page is one run against one query. Ask Eureka your own question about perceptual image quality assessment and every answer comes back with the patent numbers behind it.
Try EurekaThe most-cited records in this corpus
System and method for automated electronic catalogue management and electronic image quality assessment
The filing describes a system that receives image data for an item and runs two parallel operations: a structural similarity analysis producing a structural similarity score, and a distortion-based pipeline that generates derivative images by applying distortions, extracts features from them, and determines quality from those extracted features.Filed by Walmart Apollo, LLC — a retail-catalogue application of quality assessment rather than a codec or camera-pipeline use case.
| # | Publication no. | Patent title | Citations |
|---|---|---|---|
| 1 | US20160358321A1 | Full reference image quality assessment based on convolutional neural network | 112 |
| 2 | US20180286032A1 | Assessing quality of images or videos using a two-stage quality assessment | 41 |
| 3 | CN107633520A | 一种基于深度残差网络的超分辨率图像质量评估方法 | 41 |
| 4 | WO2018058090A1 | Method for no-reference image quality assessment | 35 |
| 5 | US20170140518A1 | System and method for comparison-based image quality assessment | 34 |
| 6 | CN101489130A | 基于图像边缘差异统计特性的全参考型图像质量评价方法 | 33 |
| 7 | CN104408707A | 一种快速数字成像模糊鉴别与复原图像质量评估方法 | 32 |
| 8 | WO2016197026A1 | Full reference image quality assessment based on convolutional neural network | 28 |
| 9 | CN108109147A | 一种模糊图像的无参考质量评价方法 | 22 |
| 10 | EP2889833A1 | Method and apparatus for image quality assessment | 20 |
Citation counts favour older filings in any searched corpus and should be read as a signal of influence on later work, not of current commercial weight.
Patent titles are shown in the language they were filed in, not translated, so that each record stays verifiable against the original filing — a translated title will not match in Eureka or in any national register. 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 numbers imply for strategy
Three findings that matter more than the raw counts on their own.
The growth phase has already passed
Peak filing activity was 2019, and the 2022 midpoint of 5 sits below it. A partial 2026 count of 2 is consistent with continued decline once the publication lag is accounted for, not with a technology still in its build-up phase.
Influence sits with a handful of early CNN-based filings
The most-cited record, a convolutional-network full-reference method, draws more than double the citations of the next entry. That concentration marks the foundational prior art examiners and drafters will keep encountering, not a current filing hotspot.
Filing is a two-country contest with a thin periphery
The United States and China together account for the large majority of receiving-office activity, with India, WIPO, Europe and Australia each in single digits. A filing strategy built only around US/CN priority will miss the PCT and India signal but won't miss much else.
Eureka can read the same corpus for gaps instead of for coverage: under-claimed branches adjacent to perceptual image quality assessment, with the prior art for and against each one.
Who holds the ground, and where it's open
No assignee in this dataset shows filing activity in the latest tracked year, including university and corporate filers that were previously active — a pattern more consistent with a maturing, quiet field than an emerging one.
Recent-year momentum has stalled across the board
Corporate filers and universities alike — from consumer electronics to retail to multiple Chinese universities — show zero filings in the most recent tracked year. That is unusual breadth for a stall, and suggests either a genuine slowdown or a publication-lag gap not yet visible in the data.
This is not a collaborative filing space
Only two co-assignee pairs appear in the entire dataset, both involving the same corporate filer paired with individual named inventors. Most families here are filed solo, which means licensing and freedom-to-operate discussions will mostly run one counterparty at a time.
Application-layer filers are a small but real minority
Beyond the core imaging classes, a handful of records extend into commerce (G06Q), signalling/alarm systems (G08B) and healthcare informatics (G16H). These are low-count today but mark where quality-assessment techniques are being repurposed outside pure imaging pipelines.
| Assignee | Recent year | YoY |
|---|---|---|
| Sony Group Corporation | 0 | — |
| Walmart Apollo, LLC | 0 | — |
| ARLO TECHNOLOGIES INC | 0 | — |
| Shenzhen University | 0 | — |
| Xi'an Jiaotong University | 0 | — |
| Huaqiao University | 0 | — |
| Shanghai Jiao Tong University | 0 | — |
| Board of Regents, The University of Texas System | 0 | — |
Where to take this
The dataset points to specific next steps depending on whether you're scoping freedom-to-operate or looking for a filing gap.
Check filing dates, not just publication dates
The zero-momentum finding across all assignees could reflect the ~18-month publication lag rather than a real stop in activity. Pull filing dates directly before concluding the field has gone cold.
Explore filing dates in EurekaMap the two-country filing pattern against your markets
With US and China accounting for most receiving-office activity and Europe notably thin, confirm whether your target markets are covered before assuming broad protection exists.
Run a geographic coverage check in EurekaTest the under-claimed branches for a first-filer position
Generative-artifact no-reference metrics and healthcare-imaging quality scoring both show thin records relative to the G06T core — worth a focused prior-art check before assuming the space is occupied.
Search white space in EurekaCommon questions about this landscape
It refers to methods that score how a human would perceive the quality or distortion of an image or video, as opposed to purely pixel-level metrics like mean squared error. Patent filings in this space cover full-reference methods (comparing a distorted image to a clean original, often with convolutional neural networks), no-reference methods (scoring an image with no original to compare against), and perceptual similarity metrics such as LPIPS or structural similarity indices. In this dataset these methods are classified mainly under image data processing (G06T), image/video recognition (G06V, G06K) and pictorial communication (H04N).
The dataset shows filing concentrated among a mix of corporate and university assignees, including consumer electronics and retail companies alongside several Chinese universities. No single assignee shows continued filing activity in the most recent tracked year, and co-assignee collaboration is minimal — only two co-filing pairs appear across all 83 families. This points to a fragmented rather than consolidated ownership picture.
Not based on this data. Filings peaked at 8 in 2019, sat at 5 by the 2022 midpoint, and show only 2 in the still-partial most recent year. Even allowing for the roughly 18-month publication lag that understates the newest years, the multi-year trend before the lag window is flat to declining rather than accelerating.
The most-cited record is a full-reference image quality assessment method built on a convolutional neural network, drawing 112 citations — more than double the next most-cited record. High citation counts in a searched corpus tend to favour older, foundational filings rather than currently active claim territory, so this signals influence on later technical approaches rather than present-day filing activity.
The core image-processing classification (G06T) is heavily claimed, but adjacent applications remain thin: healthcare-imaging quality scoring, alarm and surveillance-feed quality triggers, and no-reference metrics tailored to generative-model artifacts all show low record counts relative to the core. These are areas where a well-drafted first claim could still stake meaningful ground, though a full freedom-to-operate search against the most-cited full-reference and no-reference patents is still warranted before filing.
Research Perceptual Image Quality Assessment in depth with Eureka
Go past this page: query the whole perceptual image quality assessment 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.