Partial Motion Vector Difference Signaling for Video Bitstream Compression

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

The existing method for motion vector (MV) explicit signaling in video coding requires full information of a corresponding motion vector difference (MVD) to be signaled in a bitstream, which is costly in terms of bits.

Innovation Solution

A method where a decoder decodes partially encoded MVD information from a coded video bitstream and regenerates the remaining MVD information, allowing the encoder to only encode partial MVD information for bit-saving.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If full MVD information is signaled in the bitstream, then the motion vector can be accurately decoded, but the bit cost increases

Engineering Contradiction:

ImproveMVD decoding accuracyVSAvoidbit cost

Core Design Contradiction:

Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent extracts only the necessary sign information from the full MVD data and transmits it separately in the bitstream, while the magnitude information is derived from the MVP. This selective extraction reduces the amount of data that needs to be transmitted while preserving the ability to reconstruct the complete MVD at the decoder side.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The MVP (motion vector predictor) serves as an intermediary that provides the magnitude information for MVD, allowing the encoder to transmit only the sign information. The decoder uses this MVP as a reference to reconstruct the full MVD by combining the transmitted sign with the predicted magnitude.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Quantity of substance

If partial MVD information is encoded to reduce bit cost, then the bitstream size decreases, but the complexity of regenerating complete MVD information increases

Engineering Contradiction:

Improvebitstream sizeVSAvoiddecoder complexity

Core Design Contradiction:

Quantity of substanceVSDevice complexity

Solution Approach 1:

The encoder performs preliminary computation of the MVP (motion vector predictor) and transmits it in the bitstream along with the partial MVD sign information. This preliminary action provides the decoder with the necessary reference data to easily reconstruct the complete MVD without requiring complex search or optimization algorithms.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The decoder uses the transmitted MVP as a self-service reference to automatically derive the magnitude information needed to reconstruct the complete MVD. This self-service mechanism allows the decoder to generate the missing information using simple arithmetic operations on the received data, minimizing the need for external reference data or complex processing.

Inventive Principle:

Principle #25Self-service

Data Source

PatentUS20250142108A1Motion vector difference derivation

Publication Date: 2025.05.01 TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)

AI summary

Methods and apparatuses for processing (e.g., decoding) a current block within a current picture. The processing may include decoding partial motion vector difference (MVD) information from a coded video bitstream. The processing may include determining first and second motion vector predictors (MVPs). The processing may include determining complete MVD information using the partial MVD information and one or more of the first and second MVPs. Determining the complete MVD information may include using the partial MVD information to generate a set of MVD hypothesis pairs, and each of the MVD hypothesis pairs may include a possible first MVD and a possible second MVD. The processing may include determining first and second final prediction blocks of the current block using the complete MVD information and the first and second MVPs.