Temporal Motion Vector Predictor Shifted Position

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current video coding standards face limitations in motion information prediction, as the definition of a neighborhood for motion information prediction can be too large, leading to increased signaling costs and reduced efficiency, while being too small may not capture the best motion predictors.

Innovation Solution

The method involves identifying a collocated region in a reference image to decode a current block, obtaining a motion vector from this region, and applying a displacement to find a shifted position in another reference image, where available motion vectors are used to construct a list of predictor candidates for reconstructing the motion vector of the current block, with rescaling and clipping to ensure accurate prediction within constrained areas.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a large neighborhood is used for motion information prediction, then prediction accuracy is improved, but signaling cost increases

Engineering Contradiction:

Improvemotion information prediction accuracyVSAvoidsignaling cost

Core Design Contradiction:

Measurement precisionVSLoss of information

Solution Approach 1:

The patent segments the neighborhood into multiple candidate regions (first candidate region using first motion vector, second candidate region using second motion vector) and selects the best predictor from these segmented regions, allowing accurate prediction without signaling the entire large neighborhood

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent uses partial action by deriving motion information from a subset of candidate regions rather than the complete neighborhood, achieving sufficient prediction accuracy while reducing the amount of motion information that needs to be signaled

Inventive Principle:

Principle #16Partial or excessive action

2Loss of information

If a small neighborhood is used for motion information prediction, then signaling cost is reduced, but prediction accuracy deteriorates

Engineering Contradiction:

Improvesignaling costVSAvoidmotion information prediction accuracy

Core Design Contradiction:

Loss of informationVSMeasurement precision

Solution Approach 1:

The patent introduces a temporal dimension by using motion vectors from different reference pictures (first reference picture and second reference picture) to create multiple candidate regions, effectively expanding the search space without increasing spatial signaling cost

Inventive Principle:

Principle #17Another dimension (Dimensionality change)

3Measurement precision

If motion vectors are derived from multiple reference pictures, then prediction accuracy is improved, but device complexity increases

Engineering Contradiction:

Improvemotion information prediction accuracyVSAvoidprocessing complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent performs preliminary action by pre-defining the candidate regions and motion vector derivation rules before actual prediction, so that during decoding the processor only needs to follow predefined steps rather than performing complex real-time analysis

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS12041241B2Method and device for image encoding and decoding with a shifted position

Publication Date: 2024.07.16 INTERDIGITAL MADISON PATENT HLDG

AI summary

A method for obtaining a temporal motion vector predictor is disclosed. A new temporal motion vector predictor is obtained by using a motion vector of a block in a collocated image. The block is displaced (with respect spatial location of the current block) by a motion vector coming from a new temporal motion vector predictor (TMVP) process.