BVP Candidate Clustering for Video Encoding Accuracy

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

In video encoding and decoding systems, the accurate classification and identification of block vector predictor (BVP) candidates are hindered by non-homogeneous identification, leading to incorrect flipping types and potential pruning of incorrect BVP candidates from the candidate list, which affects the refinement and reordering process.

Innovation Solution

The proposed solution involves determining a list of BVP candidates by clustering them based on their flipping types and costs, ensuring correct classification and reordering to improve the prediction mode, specifically by indicating the flipping type of each BVP candidate in the candidate list to facilitate accurate refinement.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If BVP candidates are identified without homogeneous classification, then the candidate list includes more potential candidates, but the flipping types cannot be indicated correctly leading to incorrect identification

Engineering Contradiction:

Improvenumber of BVP candidatesVSAvoidaccuracy of flipping type identification

Core Design Contradiction:

Quantity of substanceVSMeasurement precision

Solution Approach 1:

The patent segments the BVP candidate list into multiple subsets based on flipping type characteristics. By dividing the homogeneous identification problem into smaller homogeneous groups, each subset can be accurately classified with correct flipping types indicated, resolving the contradiction between maintaining a comprehensive candidate list and achieving accurate identification.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent applies different identification and classification methods to different subsets of BVP candidates based on their local characteristics. Each subset is treated with appropriate flipping type indication rules, ensuring that the identification accuracy is optimized for each local group while maintaining the overall completeness of the candidate list.

Inventive Principle:

Principle #3Local quality

2Productivity

If BVP candidates are pruned from the candidate list, then the refinement process becomes more efficient, but incorrect BVP candidates may be removed reducing prediction accuracy

Engineering Contradiction:

Improverefinement process efficiencyVSAvoidprediction accuracy

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The patent performs preliminary classification and flipping type indication on BVP candidates before the refinement process. By pre-organizing candidates into accurate homogeneous subsets with correct flipping types identified, the subsequent refinement process can efficiently prune candidates with higher confidence, improving both efficiency and reliability without removing correct candidates.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent implements a feedback mechanism where the classification and flipping type indication results from preliminary processing guide the refinement process. The refined candidate list is fed back into the prediction process, and the results inform further refinement iterations, ensuring that only truly incorrect candidates are removed while maintaining prediction accuracy.

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS20240380895A1Clustering BVP Candidates

Publication Date: 2024.11.14 COMCAST CABLE COMM LLC

AI summary

A flipping or non-flipping type of each BVP candidate in a candidate list may require correct classification/identification, for example, for correct reordering and accurate refinement A block vector predictor (BVP) candidate list may be adjusted to provide a more accurate prediction of a block vector (BV). A coder (e.g., encoder or decoder) may determine a final candidate list, for example, by pruning invalid candidates.