Motion Vector Selection for Video Encoding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing video encoding techniques fail to accurately capture and utilize motion information from reference frames, leading to inefficiencies in inter-prediction, particularly for frames with complex non-translational motion, and do not fully exploit available motion information for improved coding efficiency.

Innovation Solution

The method involves reconstructing reference frames and projecting motion vectors onto the current frame to select a weighted motion vector for prediction, using a co-located reference frame generated through motion field estimation and refinement, which captures true motion activities and improves prediction accuracy without significant computational overhead.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If motion vectors from multiple reference frames are projected onto the current frame, then motion information accuracy is improved, but the number of projected motion vectors increases leading to increased computational complexity

Engineering Contradiction:
Improvemotion information accuracyVSAvoidcomputational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent changes the parameter of motion vector selection by introducing magnitude-based weighting. Instead of treating all projected motion vectors equally, the patent assigns weights based on vector magnitudes and selects the optimal vector through weighted comparison, thereby improving motion accuracy while managing computational complexity through parameter optimization.

Inventive Principle:
Principle #35Parameter changes

2Device complexity

If a single motion vector is selected from multiple projected motion vectors, then computational complexity is reduced, but motion information accuracy deteriorates

Engineering Contradiction:
Improvecomputational complexityVSAvoidmotion information accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent introduces magnitude as a selection parameter to differentiate between multiple projected motion vectors. By calculating weights based on vector magnitudes and selecting the vector with optimal weight, the patent achieves accurate motion representation while maintaining single-vector simplicity for prediction operations.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent implements a feedback mechanism where multiple motion vectors are projected and evaluated, their magnitudes are compared, and the optimal vector is selected based on weighted evaluation. This feedback loop ensures that the selected vector best represents the actual motion, improving accuracy while still selecting only one vector for the final prediction.

Inventive Principle:
Principle #23Feedback

3Productivity

If motion vectors are projected from both first and second reference frames, then coding efficiency is improved, but the amount of data to be processed increases

Engineering Contradiction:
Improvecoding efficiencyVSAvoiddata volume
Core Design Contradiction:
ProductivityVSQuantity of substance

Solution Approach 1:

The patent extracts only the essential motion information from multiple reference frames by projecting motion vectors and then selecting the single most representative vector based on magnitude weighting. This extraction process captures the critical motion data needed for accurate prediction while discarding redundant information, thereby improving coding efficiency without proportionally increasing data processing requirements.

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentUS20240422309A1Selection of projected motion vectors
Publication Date: 2024.12.19 GOOGLE LLC
  • US20240422309A1 patent drawing
  • US20240422309A1 patent drawing
  • US20240422309A1 patent drawing

AI summary

Methods, systems and apparatuses are disclosed including computer readable medium storing instructions used to encode or decode a video or a bitstream encodable or decodable using disclosed steps. The steps include reconstructing a first reference frame and a second reference frame for a current frame to be encoded or decoded, projecting motion vectors of the first reference frame and the second reference frame onto pixels of a current reference frame resulting in a first pixel in the current reference frame being associated with a plurality of projected motion vectors, and selecting a first projected motion vector from the plurality of projected motion vectors as a selected motion vector associated with the first pixel to be used for determining a pixel value of the first pixel, the selection based on magnitudes of the respective ones of the plurality of projected motion vectors.