L-type partitioning tree
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video encoding formats like AV1 have limited quantization step size resolution, particularly evident in 10-bit and 12-bit internal bit depth, which affects the granularity of bit rates and compression efficiency.
Innovation Solution
The proposed method introduces an extended quantizer and L-shaped partitioning for video data encoding and decoding, allowing for non-rectangular partitions in addition to rectangular ones, enhancing the partitioning flexibility and resolution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional rectangular partitioning is used in video encoding, then the encoding process is simple and fast, but the compression efficiency is limited due to fixed block shapes
Solution Approach 1:
The patent segments the video block into multiple non-rectangular partitions including L-shaped partitions, allowing the coding block to be divided into irregular regions that better match the actual content structure. This segmentation enables more precise adaptation to edges and contours in the video data, improving compression efficiency by reducing residual energy in transform coefficients.
Solution Approach 2:
The patent introduces asymmetric L-shaped partitions that break the traditional symmetric rectangular block structure. These asymmetric partitions allow one partition to extend further in certain directions, creating irregular shapes that better conform to object boundaries and motion patterns in the video content, thereby improving coding precision without requiring complex additional structures.
2Measurement precision
If quantization step size resolution is limited in AV1 format, then the encoder complexity is reduced, but the bit rate granularity and compression precision are degraded
Solution Approach 1:
The patent changes the quantization parameter resolution by introducing an extended quantizer that provides finer granularity in quantization step sizes. This allows for more precise control over the compression ratio and bit rate, enabling better optimization of compression efficiency without requiring fundamental changes to the encoder architecture.
Solution Approach 2:
The patent introduces dynamic quantization adjustment mechanisms that allow the quantization step size to vary more finely across different blocks and regions. This dynamic control enables adaptive optimization of compression performance by adjusting quantization precision locally based on content characteristics, improving overall compression efficiency while maintaining manageable encoder complexity.
3Adaptability or versatility
If non-rectangular partitions are introduced, then the partitioning flexibility and coding precision are improved, but the processing complexity increases
Solution Approach 1:
The patent segments the coding block into a specific set of non-rectangular partitions including L-shaped partitions with defined geometric relationships. By establishing clear segmentation rules and constraints on partition shapes and sizes, the patent achieves improved adaptability to content structures while keeping the processing complexity manageable through systematic partition generation algorithms.
Solution Approach 2:
The patent applies different partition shapes and sizes to different local regions of the video block based on content characteristics. L-shaped partitions are used in regions with specific edge orientations or motion patterns, while rectangular partitions are used elsewhere. This local adaptation improves coding precision without requiring complex processing across the entire block, as each region receives appropriately simplified treatment.
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.