Application Number: AU 2026201822

Transform-Based Image Coding Method, and Apparatus Therefor Smarter Transform Selection for High-Efficiency Video Compression

The invention provides image-coding and image-decoding methods that operate on transform coefficients of residual blocks, refining how the encoder selects between candidate transforms and how the decoder reconstructs the residual from the signalled transform indices. The disclosed method covers both the encoder side, which decides which transform to apply to which block based on configurable

Open for Public Inspection
AU 2026201822 Featured Image

View the Transform-Based Image Coding Method, and Apparatus Therefor PDF

Download the PDF version of this Application Open to Public Inspection

This patent describes a transform-based image and video coding method aimed at the next-generation video codecs that handle 4K, 8K and immersive content, refining how the encoder and decoder choose, signal and apply transforms to residual blocks for higher compression efficiency at the same visual quality.

The Problem

Demand for high-resolution and high-quality video keeps climbing. Streaming services, broadcast, augmented and virtual reality, and game-content distribution all push 4K and 8K content into a transmission and storage pipeline whose bit budget has not grown proportionally. Each generation of video codec, from H.264 through HEVC and into VVC and beyond, has had to find further compression efficiency without unacceptable computational overhead at the decoder. The transform stage of a codec, where spatial residual data is converted into frequency-domain coefficients before quantisation and entropy coding, is one of the principal levers. Choosing the right transform for each block, and signalling that choice to the decoder compactly, has become an increasingly subtle optimisation as the set of available transforms grows.

What This Invention Does

The invention provides image-coding and image-decoding methods that operate on transform coefficients of residual blocks, refining how the encoder selects between candidate transforms and how the decoder reconstructs the residual from the signalled transform indices. The disclosed method covers both the encoder side, which decides which transform to apply to which block based on configurable criteria, and the decoder side, which interprets the signalling to reverse the transform faithfully. The approach is positioned within the transform-based portion of modern video codecs and is compatible with related coding tools such as adaptive partitioning and intra-prediction modes.

The patent also covers the apparatus, both encoder and decoder, that performs the method, and the bitstream constructs that allow the encoder and decoder to stay in sync.

Key Features

  • Refined transform selection logic. Encoder-side selection of transform candidates per block, designed to improve coding efficiency over previous-generation codecs.
  • Compact signalling. Bitstream syntax that conveys the chosen transform to the decoder without bloating the high-level header.
  • Decoder-side reconstruction path. Decoder methods to apply the inverse transform consistent with the encoder’s choice and recover the residual.
  • Compatibility with surrounding coding tools. Integrates with intra-prediction, partitioning and quantisation strategies typical of HEVC and VVC-class codecs.
  • Encoder and decoder apparatus claims. Coverage of the hardware or software apparatus that implements the method, important for licensing in consumer electronics and silicon.

Who Is Behind It?

The applicant is LG Electronics Inc., a long-standing contributor to the standards-track video coding bodies and a major holder of essential patents in HEVC and VVC. The named inventors are Moonmo Koo, Jaehyun Lim, Seunghwan Kim and Mehdi Salehifar. The Australian application is a divisional of AU 2024203027. The Australian patent agent is Dentons in Auckland.

Why It Matters

Video compression is one of the most economically significant technical fields of the past two decades, and the patent thickets around modern codec standards are central to global consumer-electronics licensing. Australia is both a streaming market and a semiconductor design participant, and patents covering transform-stage methods that can read on standardised codec implementations have direct relevance to anyone shipping decoders into the Australian market. The Australian filing positions LG’s transform-coding IP for inclusion in essentiality declarations and for enforcement in this jurisdiction.


AU 2026201822 was published in the Australian Official Journal of Patents on 2 April 2026 and is open for public inspection. Patent applications represent inventions that are sought to be protected and do not necessarily reflect commercially available products.

Disclaimer

The information presented in this article is provided for general informational and illustrative purposes only.

Content on this page may be derived from publicly available intellectual property records, including patent documentation and related materials. While reasonable care is taken in compiling and summarising this information, ATMOSS does not guarantee the accuracy, completeness, currency, or reliability of any content presented.

This article is not a substitute for reviewing the original source documents. Patent applications, specifications, claims, and related records may contain detailed technical, legal, and contextual information that is not fully represented in this summary.


ATMOSS does not provide legal, technical, or commercial advice. Users should not rely on this content for decision-making purposes.
For authoritative and up-to-date information, users should refer directly to the official records available via IP Australia and other relevant intellectual property databases. Links to these official sources are provided where applicable.


ATMOSS accepts no liability for any loss, damage, or consequences arising from the use of, or reliance on, the information contained in this article.