Coding Unit Division Decision Method for Video Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The Top-Down approach in H.265 coding is computationally intensive, leading to low encoding speed due to the requirement of performing mode decisions strictly level by level for coding unit division decisions.
Innovation Solution
A method and device for coding unit division decision that determines whether the current level of a coding unit is smaller than a predicted level, allowing for division decisions on sub-CUs and skipping unnecessary mode decisions based on preset skip conditions, thereby reducing computational load.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the Top-Down approach is used for coding unit division decision, then the mode decision is made systematically level by level, but the encoding computational complexity increases and encoding speed decreases
Solution Approach 1:
The patent applies preliminary action by predicting the optimal level for each coding unit before performing full mode decisions. The level prediction rule pre-determines which CUs should be divided based on neighboring CU information, allowing the encoder to skip unnecessary mode decisions for CUs where the optimal division level can be predicted in advance. This reduces computational complexity while maintaining systematic decision-making.
Solution Approach 2:
The patent implements local quality by applying different processing strategies to different coding units based on their predicted characteristics. CUs with predicted levels matching current levels undergo full mode decisions, while CUs where prediction indicates optimal division skip detailed mode decisions. This localized differentiation optimizes encoding speed without compromising overall decision quality.
2Manufacturing precision
If mode decisions are performed on all CUs level by level, then complete division decisions are obtained, but the computational load increases significantly
Solution Approach 1:
The patent extracts and utilizes level information from neighboring coding units to create prediction rules. By taking out the essential level characteristics from already-processed CUs and applying them to predict future CU levels, the method avoids redundant computational operations while preserving division decision accuracy for CUs that truly require detailed analysis.
Solution Approach 2:
The patent implements skipping by allowing the encoder to bypass detailed mode decisions for coding units where the optimal division level can be accurately predicted. When the predicted level matches the current level or indicates optimal division, the encoder skips the computationally intensive mode decision process for those CUs, rushing through to the next CU while maintaining overall decision accuracy.
Data Source
AI summary
Embodiments of a coding unit division decision method and device, an encoder and a storage medium are disclosed. The method includes: determining the current level of a target CU and obtaining a predicted level of the target CU according to a preset level prediction rule; determining whether the current level is smaller than the predicted level; in case of the current level smaller than the predicted level, performing division decisions on sub-CUs of the target CU to obtain first division decision results; in case of the first division decision results meeting a preset first skip condition, determining that a division decision result for the target CU is a result comprised of a division result for dividing the target CU into the four sub-CUs and the first division decision results respectively corresponding to the four sub-CUs of the target CU.