Motion Vector Refinement for Video Compression

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current video coding technologies face challenges in efficiently compressing digital video due to increasing demand for higher resolution video, leading to higher bandwidth requirements.

Innovation Solution

The proposed solution involves a video processing method that uses motion refinement based on updated motion vectors generated through a two-step inter-prediction process, applicable to existing and future video coding standards like HEVC.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional motion prediction methods are used, then device complexity is reduced, but measurement precision of motion vectors deteriorates

Engineering Contradiction:
Improvemotion vector precisionVSAvoidprocessing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The motion vector processing is divided into two distinct stages: first generating initial motion vectors from reference blocks, then refining them through a second prediction step. This segmentation allows each stage to focus on specific aspects of motion compensation, achieving higher overall precision without proportionally increasing complexity.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary motion vector generation using reference blocks before the final prediction step. By pre-calculating motion vectors from available reference data and then refining them, the system achieves higher precision while managing complexity through staged processing.

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If higher resolution video is transmitted, then video quality is improved, but bandwidth consumption increases

Engineering Contradiction:
Improvevideo qualityVSAvoidbandwidth
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent uses feedback mechanisms where prediction errors from the first stage inform the second prediction stage. By continuously refining motion vectors based on residual errors and using multiple reference blocks with different prediction modes, the system achieves better compression efficiency for high-resolution video, reducing the bandwidth required to maintain video quality.

Inventive Principle:
Principle #23Feedback

3Measurement precision

If motion vector refinement is performed, then prediction accuracy is improved, but processing time increases

Engineering Contradiction:
Improveprediction accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The refinement process is segmented into two distinct prediction passes rather than one exhaustive pass. The first pass generates initial motion vectors using available reference blocks, and the second pass refines them. This segmentation reduces the computational burden of each individual pass while achieving cumulative improvement in accuracy.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies partial refinement by using multiple reference blocks selectively rather than exhaustively searching all possible references. By combining results from multiple partial predictions with different reference blocks, the system achieves near-optimal accuracy with reduced processing time compared to exhaustive search methods.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentUS20250039434A1Constraints for usage of updated motion information
Publication Date: 2025.01.30 DOUYIN VISION CO LTD
  • US20250039434A1 patent drawing
  • US20250039434A1 patent drawing
  • US20250039434A1 patent drawing

AI summary

Embodiments of the present disclosure relates to constraints for usage of updated motion information. A method for video processing is provided, including: determining original motion information associated with a current block; generating updated motion information based on specific prediction mode; and performing, based on the updated motion information, a conversion between the current block and a bitstream representation of a video data including the current block, wherein the specific prediction mode includes one or more of bi-directional optical flow (BIO) refinement, a decoder-side motion vector refinement (DMVR), frame-rate up conversion (FRUC) techniques or a template matching technique.