Samsung · Filed Feb 27, 2026 · Published Jul 9, 2026 · verified — real USPTO data

Samsung Patents a Lookup-Table Filter That Cleans Up Compressed Video Frames

Every time a video is compressed, the image loses a little fidelity, and fixing those artifacts after the fact is harder than it sounds. Samsung is patenting a method that uses pre-built lookup tables to quickly correct individual pixels at the moment a video is decoded.

Samsung Patent: Lookup-Table Image Filtering for Video Codecs — figure from US 2026/0197448 A1
Figure from the official USPTO publication.
Publication number US 2026/0197448 A1
Applicant SAMSUNG ELECTRONICS CO., LTD.
Filing date Feb 27, 2026
Publication date Jul 9, 2026
Inventors Hyungmin ROH, Jeeyoon PARK
CPC classification 375/240.02
Grant likelihood Medium
Examiner CENTRAL, DOCKET (Art Unit OPAP)
Status Docketed New Case - Ready for Examination (Apr 1, 2026)
Parent application is a Continuation of PCTKR2024012148 (filed 2024-08-14)
Document 20 claims

How Samsung's filter smooths out compressed video pixels

Imagine streaming a movie and noticing soft, blocky edges around a character's face. That's compression doing its job, squishing video data down to a manageable size, but leaving behind visual leftovers. Samsung's patent describes a smarter cleanup step that happens after the video is uncompressed on your device.

The idea is to look at each pixel in a reconstructed image, compare it to its neighbors, and use a pre-built reference table to decide whether and how much to adjust its brightness. The table is chosen based on how the video was predicted and encoded in the first place, so the correction is tailored rather than one-size-fits-all.

The result is a filtered, cleaner pixel with less of the ringing and blockiness that compressed video often produces. Because the lookup tables are pre-computed, the process can run fast without heavy calculation on the fly, which matters a lot on phones and TVs where battery and processing power are limited.

How the LUT maps pixel differences to correction values

The patent describes a post-reconstruction filtering step inside a video codec (the software that encodes and decodes video). After a block of pixels is decoded (called a reconstructed block), the method picks a center sample (one pixel) and measures its brightness difference from each of its neighbors.

Those differences are fed into a lookup table (LUT), essentially a pre-calculated cheat sheet that maps each difference value to a small correction number called a modifier value. Crucially, the correct LUT is chosen based on the block's prediction mode (the encoding strategy used when the video was originally compressed), so the filter adapts to how that part of the image was originally handled.

  • All the modifier values are summed into a single filtering parameter.
  • That parameter maps to an offset value, how much to nudge the center pixel's brightness up or down.
  • The final filtered sample is the original pixel plus that offset.

This is all part of the in-loop filtering stage used in modern video codecs like HEVC, VVC, and AV1. By tying the LUT choice to the prediction mode, Samsung's approach aims to make corrections that are aware of the local image structure rather than treating every pixel the same way.

What this means for video streaming and device efficiency

For everyday users, this kind of filtering is what separates a 4K stream that looks crisp on a large TV from one that shows blocky artifacts around fast-moving objects or high-contrast edges. Better in-loop filtering directly improves perceptual video quality without requiring more bandwidth or a higher bitrate file.

The LUT-based design is also computationally cheap compared to neural-network approaches, which matters for mobile devices and smart TVs that decode video in real time on constrained hardware. Samsung produces both the chips inside televisions and phones and the codec software running on them, so this kind of filtering improvement could show up across a wide range of its own products.

Editorial take

This is a workmanlike but genuinely useful codec patent. Post-reconstruction filtering is a well-established battleground in video compression standards, and Samsung's LUT-based approach is interesting precisely because it stays lightweight. It's not flashy research; it's the kind of incremental quality improvement that actually ships in televisions and phones.

Which company should we read for you?

We track 17 companies here. Pro is the same weekly breakdown for any company you choose, delivered privately. Type a name and we'll scope it and send you a quote.

Get one Big Tech patent every Sunday

Plain English, intelligent commentary, no hype. Free.

Source. Full patent text and figures from the official USPTO publication PDF.

Editorial commentary on a publicly published patent application. Not legal advice.