Adaptive Block Partitioning for Lower-Complexity Video Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The complexity and processing amount increase when image blocks are split into non-rectangular partitions during video encoding and decoding, leading to inefficiencies.
Innovation Solution
An encoder and decoder that determine whether to split image blocks into non-rectangular partitions based on size parameters and thresholds, allowing efficient processing by optionally omitting the split parameter or using rectangular partitions when certain conditions are met.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If image blocks are split into non-rectangular partitions, then encoding precision and adaptability improve, but device complexity and processing amount increase
Solution Approach 1:
The patent applies parameter changes by dynamically adjusting the partition shape parameter (rectangular vs. non-rectangular) based on image block characteristics. When complexity thresholds are exceeded, the system transitions from non-rectangular to rectangular partitions, effectively using parameter variation to balance precision and complexity requirements.
Solution Approach 2:
The patent implements dynamics by making the partition shape adaptive rather than static. The encoder dynamically determines whether to use rectangular or non-rectangular partitions based on real-time assessment of image block properties and complexity metrics, allowing the system to optimize between precision and processing load adaptively.
2Manufacturing precision
If image blocks are split into non-rectangular partitions, then encoding precision improves, but productivity decreases
Solution Approach 1:
The system changes the partition shape parameter based on complexity assessments, switching between rectangular and non-rectangular formats. This parameter adjustment ensures high encoding precision for suitable blocks while maintaining processing efficiency by using simpler rectangular partitions when complexity is high.
Solution Approach 2:
The patent applies partial action by using non-rectangular partitions only for image blocks that meet specific complexity criteria, rather than applying them universally. This selective approach maintains encoding precision where needed while avoiding the productivity penalty of processing all blocks with complex partitions.
3Measurement precision
If motion vector candidate lists include sub-partition motion vectors for non-rectangular partitions, then measurement precision improves, but device complexity increases
Solution Approach 1:
The patent applies local quality by differentiating the treatment of motion vector candidates based on partition type. For non-rectangular partitions, the system selectively includes or excludes sub-partition motion vectors based on local complexity characteristics, rather than applying a uniform approach across all partitions.
Solution Approach 2:
The system changes the composition parameter of motion vector candidate lists dynamically. When non-rectangular partitions are detected with high complexity, the system adjusts the candidate list parameters to exclude sub-partition motion vectors, thereby reducing complexity while maintaining precision where it matters most.
Data Source
AI summary
An encoder, when sub-block encoding is to be performed, determines a plurality of sub-blocks in a first image block, the plurality of sub-blocks including a first sub-block, determines a first motion vector for the first sub-block by referring to a first candidate list, performs first inter prediction processing on the first sub-block using the first motion vector, and encodes the first image block using a result of the first inter prediction processing. When partition encoding is to be performed, the encoder, in operation, determines a plurality of partitions in a second image block, the plurality of partitions including a first partition, determines a second motion vector for the first partition by referring to a second candidate list, performs second inter prediction processing on the first partition using the second motion vector, and encodes the second image block using a result of the second inter prediction processing.