Image Encoding and Decoding With Candidate-List Prediction Refinement

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image compression technologies face challenges in achieving high accuracy and efficiency, particularly in handling high-resolution and high-quality images, due to limitations in inter prediction methods, especially in multi-hypothesis prediction modes.

Innovation Solution

The method involves bidirectional prediction to generate a basic prediction block, configures a candidate list for multi-hypothesis prediction, derives motion information, and performs a weighted sum to refine the prediction block, while reordering and refining candidate lists based on cost calculations using SAD or SATD differences.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If multi-hypothesis prediction mode is used to improve prediction accuracy, then prediction accuracy is improved, but device complexity increases due to multiple candidate lists (motion information, weight, interpolation filter)

Engineering Contradiction:
Improveprediction accuracyVSAvoidcandidate list management complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent applies preliminary action by pre-configuring multiple candidate lists (motion information candidate list, weight candidate list, interpolation filter candidate list) before the prediction process. These candidate lists are prepared in advance and stored, allowing the prediction system to quickly select and switch between different hypotheses without complex real-time computations, thus reducing operational complexity while maintaining high prediction accuracy.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent segments the prediction system into multiple independent candidate lists, each containing different hypotheses for motion information, weight, and interpolation filter. This segmentation allows the system to manage complexity by dividing the prediction task into separate, manageable components that can be independently optimized and selected, rather than handling all prediction parameters as a single complex unit.

Inventive Principle:
Principle #1Segmentation

2Measurement precision

If multiple candidate lists are configured for multi-hypothesis prediction, then prediction accuracy is improved, but signaling bits increase due to additional information required to indicate selected candidates

Engineering Contradiction:
Improveprediction accuracyVSAvoidsignaling bits
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The patent extracts and separates the candidate list configuration from the main prediction data, allowing the signaling information to be extracted as separate metadata. This enables efficient encoding where only the indices or selections from the pre-configured candidate lists need to be signaled, rather than transmitting all candidate information, thus reducing the signaling bit requirement while maintaining prediction accuracy.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent uses copying by referring to pre-configured candidate lists through indices or references. Instead of transmitting complete candidate list data, the system transmits references (indices) to the pre-stored candidate lists, and the actual candidate data is copied from the reference locations during decoding. This significantly reduces the amount of signaling bits needed while preserving all necessary prediction information.

Inventive Principle:
Principle #26Copying

3Productivity

If candidate lists are reordered based on cost calculation using SAD or SATD, then compression efficiency is improved, but calculation time increases due to additional cost computations

Engineering Contradiction:
Improvecompression efficiencyVSAvoidcalculation time
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The patent applies preliminary action by pre-calculating and storing cost values (SAD or SATD) for multiple candidates during the encoding phase, so that during decoding, the system can directly use these pre-computed costs to determine the best prediction hypothesis without performing time-consuming recalculations. This shifts the computational burden to the encoding phase where time is less critical, thereby improving decoding speed and overall compression efficiency.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent applies partial action by calculating costs for only the necessary number of candidates required for the prediction process, rather than exhaustively evaluating all possible hypotheses. The system computes costs for a limited set of top candidates that are most likely to provide accurate predictions, achieving sufficient compression efficiency without the time penalty of complete exhaustive search.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP4598016A1Image encoding/decoding method and device, and recording medium on which bitstream is stored
Publication Date: 2025.08.06 LG ELECTRONICS INC
  • EP4598016A1 patent drawingFigure 1
  • EP4598016A1 patent drawingFigure 2
  • EP4598016A1 patent drawingFigure 3

AI summary

An image decoding/encoding method and device according to the present disclosure perform bidirectional prediction so as to generate a basic prediction block of the current block, configure a candidate list for multi-hypothesis prediction, derive motion information about the current block on the basis of the candidate list, generate an additional prediction block of the current block on the basis of the motion information, and obtain the weighted sum of the basic prediction block and the additional prediction block so as to generate a final prediction block of the current block.