Video Signal Encoding and Decoding Method, and Apparatus Therefor Merge Motion Difference Coding for Next-Generation Video Compression

This patent covers a video coding technique known as merge motion difference coding, in which a small offset vector is added to a motion vector predicted from a neighbouring block, and the magnitude of that offset is signalled with a precision-aware index, providing finer motion representation while keeping the bitstream compact.

The Problem

Display panels keep getting larger and higher resolution, and consumers increasingly expect 4K and 8K streaming, AR/VR video, and high frame rate content on phones, tablets, and televisions. Each step up in pixel count multiplies the raw data rate. The compression standards that make these services viable, including H.264/AVC and High Efficiency Video Coding (HEVC, finalised in 2013), have been pushed close to their efficiency limits. The successor standards, including Versatile Video Coding (VVC, 2020), need to halve the bitrate again at the same visual quality. A large share of bitstream cost in modern coders goes to motion information: for each block, the encoder must transmit motion vectors that describe how the block has moved relative to a reference frame. Standard “merge” mode reuses a motion vector from a neighbouring block, which is cheap to signal but coarse. Predicting plus full residual signalling is accurate but expensive. The compression problem is to bridge the two: get more accurate motion than basic merge without sending a full motion vector difference for every block.

What This Invention Does

The invention provides a video signal encoding and decoding method that introduces a merge motion difference coding mode. When the mode is applied to a block, the decoder takes a merge candidate motion vector and adds an offset vector to produce the final motion vector for the block. The offset vector is described by a direction and a magnitude that are signalled with index information. Critically, the magnitude is selected from a set of motion magnitude candidates whose maximum and minimum values are set differently depending on the motion vector precision of the current block. When the maximum number of merge candidates that the merge list may include is more than one, an index is encoded to pick the candidate; when the maximum is exactly one, that index is omitted entirely, saving bits. The invention also covers the matching method on the encoder side, the bitstream that carries the data, and a non-transitory computer-readable medium storing the program and bitstream, the standard claim set required for adoption in international video coding standards.

Key Features

• Offset-on-merge motion model. Rather than choose between cheap-but-coarse merge and expensive-but-accurate motion vector difference coding, the block uses a merge candidate plus a small additional offset, balancing bit cost and accuracy.
• Precision-adaptive magnitude set. The maximum and minimum offset magnitudes change with the motion vector precision in use (for example quarter-pel, half-pel, integer-pel, four-pel), so the offset set matches the granularity actually needed.
• Conditional index encoding. When only one merge candidate is permitted, the index is not transmitted at all. This avoids wasted bits in a common edge case where the list collapses to a single candidate.
• Bitstream and storage medium claims. The patent covers not only the encoder and decoder methods but also the bitstream itself and a non-transitory storage medium storing a computer program and bitstream, the form of claim adopted for standards-essential video coding patents.
• Divisional pedigree. The application is the second-generation divisional in a family rooted in PCT/KR2019/015198, the kind of progressive division used to maintain examination scope on multiple aspects of a single specification.

Who Is Behind It?

The applicant is Guangdong Oppo Mobile Telecommunications Corp., Ltd., the Dongguan-based smartphone and consumer electronics company that has been an active contributor to video coding standards through its R&D arm. The named inventor is Bae Keun Lee. The Australian patent attorney is Spruson and Ferguson in Sydney. The application is a divisional of AU 2024278452 (filed 16 December 2024), which is itself a divisional of AU 2019376595, the national phase of PCT/KR2019/015198 filed 23 April 2021.

Why It Matters

Smartphone makers and platform owners pay or receive substantial royalty flows under the patent pools that cover modern video codecs (HEVC, VVC, AV1 disputes aside). Oppo is one of the largest smartphone shippers worldwide, with material market share in Australia through its own brand and through OnePlus. Holding Australian patent rights to fine-grained motion coding techniques that may be incorporated into VVC or future codecs strengthens Oppo’s position in cross-licensing negotiations, in standards pool participation, and in defending its products against infringement claims by competing pool members. For Australian network operators, broadcasters, and OTT video services, the practical impact is that the codecs they license to compress 4K and 8K streams contain techniques covered by IP held by handset manufacturers, not only by the traditional video patent holders.

Related Concepts

Merge motion difference coding is a refinement of motion compensation, the technique by which video codecs avoid sending whole frames again when a block has merely shifted between inter frames. It sits alongside transform and entropy coding as one of the major bit-saving levers in modern VVC and successor designs.

Patents on these primitives become commercially significant once a tool is written into a published standard, at which point they may qualify as standards-essential patents and enter the codec licensing pools.

---

AU 2026201848 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.