Reference Block Motion Search With Parallel Fractional Sample Evaluation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing video encoding techniques require iterative and multi-stage motion searches for fractional samples, leading to increased processing time and difficulty in hardware implementation due to dependency and delay in evaluating samples at different resolutions.

Innovation Solution

A single-stage fractional search method that determines fractional samples at different pel resolutions concurrently, allowing for parallel evaluation of cost metrics to identify the best reference block, reducing processing time and hardware complexity.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If iterative multi-stage motion search is used for fractional samples, then measurement precision of reference block is improved, but processing time increases and device complexity increases

Engineering Contradiction:

Improvereference block identification accuracyVSAvoidprocessing time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent combines multiple motion search stages (integer sample search, half-pel search, quarter-pel search) into a single unified stage. All fractional samples at different resolutions are evaluated concurrently in one processing pass, eliminating the sequential iteration between stages and reducing processing time while maintaining reference block identification accuracy.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The patent segments the motion search process by identifying and evaluating fractional samples at different resolutions (half-pel, quarter-pel) independently within a single stage. This segmentation allows parallel evaluation of cost metrics for samples at different resolutions without requiring sequential multi-stage processing.

Inventive Principle:

Principle #1Segmentation

2Measurement precision

If iterative multi-stage motion search is used for fractional samples, then measurement precision of reference block is improved, but device complexity increases

Engineering Contradiction:

Improvereference block identification accuracyVSAvoidhardware implementation complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent merges multiple motion search stages into a single unified stage, reducing hardware complexity by eliminating the need for multiple sequential processing pipelines. The unified stage processes all fractional samples at different resolutions concurrently, simplifying hardware architecture while maintaining measurement precision.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The patent segments the evaluation of fractional samples by resolution, allowing independent parallel processing of half-pel and quarter-pel samples within a single stage. This segmentation reduces device complexity by avoiding the inter-stage dependencies and synchronization requirements of iterative multi-stage approaches.

Inventive Principle:

Principle #1Segmentation

3Loss of time

If single-stage fractional search is used, then processing time is reduced, but measurement precision may be compromised

Engineering Contradiction:

Improveprocessing timeVSAvoidreference block identification accuracy

Core Design Contradiction:

Loss of timeVSMeasurement precision

Solution Approach 1:

The patent segments fractional samples by resolution (half-pel, quarter-pel) and evaluates cost metrics for each segment independently within the single stage. This segmentation ensures that measurement precision is maintained for each resolution level while enabling concurrent processing that reduces overall processing time.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent changes the processing parameter from sequential multi-stage evaluation to concurrent single-stage evaluation of fractional samples. By evaluating all fractional samples at different resolutions simultaneously with appropriate cost metric calculations, the patent maintains measurement precision while reducing processing time.

Inventive Principle:

Principle #35Parameter changes

4Productivity

If single-stage fractional search is used, then productivity is improved, but device complexity decreases

Engineering Contradiction:

Improvevideo encoding throughputVSAvoidhardware complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The patent merges multiple motion search stages into a single unified stage, improving productivity by enabling concurrent processing of all fractional samples. This merging simultaneously reduces device complexity by eliminating inter-stage dependencies and synchronization overhead while increasing video encoding throughput.

Inventive Principle:

Principle #5Merging (Combining)

Data Source

PatentUS12556709B2Motion search for reference block selection

Publication Date: 2026.02.17 QUALCOMM INC

AI summary

A method of encoding video data includes determining an integer sample in a reference picture of the video data; determining, based on the integer sample, at least a first fractional sample and a second fractional sample, wherein the first fractional sample has a first fractional pel resolution, and the second fractional sample has a second fractional pel resolution different from the first fractional pel resolution; subsequent to determining both the first fractional sample and the second fractional sample, determining a first cost metric associated with the first fractional sample and a second cost metric associated with the second fractional sample; determining a reference block for a current block based on at least one of the first cost metric or the second cost metric; and encoding the current block based on the reference block.