Decoder-Side Motion Vector Refinement With Early Search Termination

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The increasing demand for high-resolution, high-quality video and image content, including virtual reality and augmented reality, requires more efficient image compression techniques to reduce transmission and storage costs while maintaining quality.

Innovation Solution

Implementing decoder-side motion vector refinement (DMVR) to derive refined motion vectors for improved inter coding efficiency, using search termination conditions to enhance image coding methods.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If decoder-side motion vector refinement (DMVR) is applied to improve inter coding efficiency, then compression efficiency is improved, but processing complexity increases

Engineering Contradiction:
Improvecompression efficiencyVSAvoidprocessing complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent applies partial action by implementing early termination conditions in the DMVR process. Instead of performing the complete refinement search for all blocks, the algorithm terminates the search early when certain conditions are met (such as when the motion vector difference becomes sufficiently small or when a maximum iteration count is reached). This partial execution of the refinement process reduces processing complexity while maintaining the primary benefit of improved compression efficiency for blocks that require refinement.

Inventive Principle:
Principle #16Partial or excessive action

2Manufacturing precision

If high resolution, high quality video/image is transmitted or stored, then image quality is improved, but transmission and storage costs increase

Engineering Contradiction:
Improveimage qualityVSAvoidtransmission and storage costs
Core Design Contradiction:
Manufacturing precisionVSQuantity of substance

Solution Approach 1:

The patent changes the parameter of motion vector precision through the DMVR process. By refining motion vectors to sub-pixel accuracy and applying iterative optimization, the system achieves better prediction accuracy for high-resolution video. This parameter change in motion vector precision enables more efficient compression of high-quality video, reducing the bitrate required to maintain a given image quality level, thereby lowering transmission and storage costs.

Inventive Principle:
Principle #35Parameter changes

3Device complexity

If search termination condition is applied in decoder-side motion vector refinement, then processing complexity is reduced, but refinement accuracy may be compromised

Engineering Contradiction:
Improveprocessing complexityVSAvoidrefinement accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent implements feedback mechanisms through iterative refinement loops in the DMVR process. The algorithm continuously evaluates the motion vector refinement quality against termination conditions (such as motion vector difference thresholds or cost function improvements). This feedback loop allows the system to adaptively determine when sufficient refinement has been achieved, balancing processing complexity and refinement accuracy by terminating early only when quality requirements are met.

Inventive Principle:
Principle #23Feedback

Data Source

PatentUS12542906B2Method and device for image coding based on decoder-side motion vector refinement
Publication Date: 2026.02.03 LG ELECTRONICS INC
  • US12542906B2 patent drawing
  • US12542906B2 patent drawing
  • US12542906B2 patent drawing

AI summary

An image decoding method performed by a decoding device is provided. The method comprises the steps of: deriving a motion vector of a current block; deriving a refined motion vector regarding the current block by applying decoder-side motion vector refinement (DMVR) to the motion vector; deriving prediction samples regarding the current block on the basis of the refined motion vector; and generating reconstruction samples regarding the current block on the basis of the prediction samples, wherein the refined motion vector is derived on the basis of a search end condition of the decoder-side motion vector refinement.