Inter-Frame Prediction Using Neighbor-Guided GPM Split Modes

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current inter-frame prediction methods in video coding, such as those using Geometric Partitioning Mode (GPM), suffer from high computing complexity and transmission overheads due to traversing a list of 64 preset modes to determine the split mode for a coding block.

Innovation Solution

An inter-frame prediction method that utilizes edge continuation information by determining the geometric partitioning mode parameter of a current coding block based on a target coding block, reducing complexity by constructing a split mode candidate list from neighboring blocks and performing weighted fusion on predicted values.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a list of 64 preset modes is traversed to determine a GPM split mode, then split accuracy is improved, but computing complexity increases

Engineering Contradiction:
Improvesplit accuracyVSAvoidcomputing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent performs preliminary action by determining the geometric partitioning mode parameter of the current coding block based on the parameter of a target coding block before final prediction. This preliminary determination narrows down the search space and avoids traversing all 64 preset modes, thus reducing computing complexity while maintaining split accuracy through edge continuation information from neighboring blocks.

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If a list of 64 preset modes is traversed to determine a GPM split mode, then split accuracy is improved, but transmission overheads increase

Engineering Contradiction:
Improvesplit accuracyVSAvoidtransmission overheads
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The patent extracts and utilizes edge continuation information from neighboring coding blocks to predict the geometric partitioning mode parameter of the current coding block. By taking out and reusing this information from surrounding blocks, the method avoids transmitting extensive mode information while maintaining accurate split decisions, thereby reducing transmission overheads.

Inventive Principle:
Principle #2Taking out (Extraction)

3Loss of information

If geometric partitioning mode parameter is determined based on target coding block, then transmission overheads are reduced, but split accuracy may deteriorate

Engineering Contradiction:
Improvetransmission overheadsVSAvoidsplit accuracy
Core Design Contradiction:
Loss of informationVSMeasurement precision

Solution Approach 1:

The patent applies feedback by performing weighted fusion of the first predicted value and second predicted value of the current coding block based on the determined geometric partitioning mode parameter. This feedback mechanism combines multiple prediction sources, allowing the system to maintain high split accuracy even when using compressed parameter representation, thus preventing accuracy deterioration despite reduced transmission overheads.

Inventive Principle:
Principle #23Feedback

Data Source

PatentUS20260082029A1Inter-frame prediction method, decoding method, electronic device, and storage medium
Publication Date: 2026.03.19 ZTE CORP
  • US20260082029A1 patent drawing
  • US20260082029A1 patent drawing
  • US20260082029A1 patent drawing

AI summary

An inter-frame prediction method includes acquiring a current video frame and determining a current coding block in the current video frame; determining a first predicted value and a second predicted value of the current coding block; determining a geometric partitioning mode parameter of the current coding block according to a geometric partitioning mode parameter of a target coding block corresponding to the current coding block; and performing a weighted fusion on the first predicted value and the second predicted value according to the geometric partitioning mode parameter of the current coding block to obtain a predicted value of the current coding block.