L-Shaped Partitioning Tree for Video Block Coding Bit Rate Granularity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing video coding formats, such as AV1, have limited quantization step size resolution, particularly for 10-bit and 12-bit internal bit depth, which affects the granularity of bit rates and efficiency of video compression.
Innovation Solution
Implementing an extended quantization step size resolution by using an L-shaped partitioning tree method that partitions video data into rectangular and non-rectangular blocks, including L-shaped partitions, to enhance the flexibility and efficiency of video encoding and decoding processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional rectangular partitioning is used in video coding, then the coding structure is simple and easy to implement, but the bit rate granularity is coarse and compression efficiency is limited
Solution Approach 1:
The patent applies segmentation by dividing the video block into multiple rectangular partitions arranged in a grid structure, where each partition can be independently coded. This allows finer control over bit rate allocation across different regions of the video block, improving bit rate granularity while maintaining a systematic and manageable coding structure.
Solution Approach 2:
The patent extends the traditional single-level partitioning by introducing a multi-level hierarchical partitioning structure. This adds a dimensional aspect to the partitioning, allowing both coarse and fine granularity control at different levels of the hierarchy, thereby improving bit rate precision without excessively increasing complexity.
2Adaptability or versatility
If fixed block partitioning is used, then the encoding process is simple and fast, but the adaptability to different video content characteristics is poor
Solution Approach 1:
The patent implements dynamic partitioning where the number, size, and arrangement of rectangular partitions are adaptively determined based on video content characteristics such as motion complexity and texture patterns. This allows the encoding structure to dynamically adjust to different content types, improving adaptability while the patent also provides default configurations to maintain reasonable encoding speed.
Solution Approach 2:
The patent applies local quality by allowing different partitions to have different coding parameters and characteristics tailored to their local content properties. Regions with high motion or complex texture can use finer partitioning, while smooth regions use coarser partitioning, optimizing both adaptability and processing efficiency.
3Productivity
If non-rectangular partitions are introduced to improve compression efficiency, then the compression performance increases, but the complexity of partitioning structure and processing increases significantly
Solution Approach 1:
The patent segments the video block into multiple rectangular partitions that can be independently processed. This segmentation approach maintains structural simplicity while achieving improved compression efficiency through selective application of different coding techniques to different partitions, avoiding the complexity of non-rectangular partition definitions.
Solution Approach 2:
The patent introduces a multi-level hierarchical dimension to the partitioning structure, allowing rectangular partitions to be recursively subdivided. This provides flexible complexity control where higher levels offer coarse partitioning for efficiency and lower levels provide fine-grained control where needed, achieving good compression without requiring complex non-rectangular shapes.
Data Source
AI summary
An example method of video decoding includes receiving a video bitstream comprising a plurality of blocks, including a current block. The method also includes performing a first partitioning of the current block such that the current block is partitioned into a first rectangular partition and an L-shaped partition, where the first rectangular partition has a same aspect ratio as the current block, and after performing the first partitioning, partitioning the L-shaped partition into a plurality of second rectangular partitions, where the plurality of second rectangular partitions includes partitions having different aspect ratios. The method further includes decoding the current block according to the first rectangular partition and the plurality of second rectangular partitions.