L-type partitioning tree
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video encoding formats, such as AV1, have limited quantization step size resolution, particularly for 10-bit and 12-bit internal bit depths, which affects the granularity of bit rates and compression efficiency.
Innovation Solution
The proposed method introduces an extended quantizer for video encoding and decoding, allowing for non-rectangular partitioning, including L-shaped partitions, to improve the resolution of quantization step sizes and enhance compression efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional rectangular partitioning is used in video encoding, then the encoding process is simple and straightforward, but the quantization step size resolution is limited and compression efficiency is reduced
Solution Approach 1:
The patent applies segmentation by dividing the coding block into multiple non-rectangular partitions (L-shaped, T-shaped, cross-shaped) instead of conventional rectangular partitions. This allows different quantization step sizes to be applied to different partition types, thereby increasing quantization resolution and compression efficiency while managing complexity through systematic classification of partition shapes.
Solution Approach 2:
The patent introduces asymmetric partitioning structures (L-shaped, T-shaped, cross-shaped partitions) that deviate from the symmetric rectangular partitions used in conventional video encoding. These asymmetric shapes enable more flexible adaptation to video content characteristics, improving quantization precision and compression efficiency by better matching the actual content distribution patterns.
2Productivity
If non-rectangular partitioning is introduced to improve compression efficiency, then quantization step size resolution increases, but the encoding and decoding process becomes more complex
Solution Approach 1:
The patent applies local quality by assigning different quantization step sizes to different partition shapes based on their local content characteristics. Each partition type (L-shaped, T-shaped, cross-shaped, rectangular) can have its own optimized quantization parameters, allowing the encoder to adapt locally to content variations and improve compression efficiency without requiring complete re-encoding of the entire block.
Solution Approach 2:
The patent introduces dynamic partitioning where the partition shape and quantization parameters can vary across different coding blocks based on content analysis. The encoder dynamically selects among multiple partition types (L-shaped, T-shaped, cross-shaped, rectangular) and adjusts quantization step sizes accordingly, enabling adaptive compression that responds to local content characteristics while maintaining manageable complexity through predefined partition templates.
Data Source
AI summary
An approach for encoding/decoding video data including obtaining video data; parsing the obtained video data to obtain partitioning parameters; obtaining a partitioned coding block based on the partitioning parameters; and decoding the obtained video data based on the partitioned coding block. The partitioned coding block may be partitioned into a rectangular partition and at least one non-rectangular partition.