Multitree Subdivision for Video Coding Adaptivity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image and video coding standards are limited in their ability to sub-divide pictures into blocks, leading to inefficient use of side information for signaling prediction parameters, especially for arbitrarily shaped objects, which increases encoding complexity and reduces coding efficiency.
Innovation Solution
A multitree subdivision method is employed to divide an array of information samples into tree root regions, allowing for recursive multi-partitioning into smaller simply connected regions, with shared coding parameters among collocated leaf blocks, and using depth-first traversal for efficient signaling and reconstruction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a picture is divided into many small blocks to adapt to arbitrarily shaped objects, then the adaptivity to picture content is improved, but the side information rate for signaling prediction parameters increases significantly
Solution Approach 1:
The patent applies segmentation by dividing the picture into a hierarchical structure of blocks at multiple levels (e.g., 16x16 macroblocks, 8x8 blocks, 4x4 blocks). This allows the system to adapt to arbitrarily shaped objects by selecting appropriate block sizes while reducing the need to signal parameters for every small block independently, as coarser blocks can share parameters with finer subdivisions.
Solution Approach 2:
The patent introduces a hierarchical dimension to the block structure, organizing blocks into multiple levels of abstraction. Instead of a flat structure where all blocks are at the same granularity, the hierarchical arrangement allows parameters to be defined at parent levels and inherited or refined at child levels, reducing the total number of parameters that need explicit signaling.
2Productivity
If the block subdivision possibilities are increased to better represent objects, then the coding efficiency is improved, but the encoding complexity increases
Solution Approach 1:
The patent implements dynamic block subdivision where the partitioning structure is not fixed but adapts based on picture content characteristics. The encoder can dynamically select which blocks to subdivide and to what extent, allowing flexible adaptation to different scene complexities while maintaining manageable encoding complexity through selective refinement rather than uniform fine-grained subdivision.
Solution Approach 2:
The patent applies local quality by allowing different regions of the picture to have different block subdivision levels. Complex regions with arbitrary shapes can be subdivided into smaller blocks for better representation, while simpler regions can use coarser blocks. This localized adaptation improves coding efficiency where needed without unnecessarily increasing encoding complexity across the entire picture.
3Loss of information
If larger block sizes are used to reduce side information rate, then the bit rate for prediction parameters is reduced, but the adaptivity to object boundaries deteriorates
Solution Approach 1:
The patent employs a nested block structure where smaller blocks are contained within larger blocks in a hierarchical manner. Larger blocks provide a coarse representation that keeps the side information rate low, while smaller blocks are nested within to capture fine details and object boundaries. This nested arrangement allows the system to maintain low overhead while preserving adaptivity where necessary.
Solution Approach 2:
The patent uses preliminary action by first establishing a coarse block structure with larger blocks that define the overall picture partitioning. This preliminary structure provides a baseline representation with low side information requirements. Subsequently, selective refinement is applied by subdividing only those blocks that require finer detail, rather than starting with fine-grained blocks and merging them.
Data Source
Figure 1
Figure 2
Figure 3A~3C
AI summary
A better compromise between encoding complexity and achievable rate distortion ratio, and/or to achieve a better rate distortion ratio is achieved by using multitree sub-divisioning not only in order to subdivide a continuous area, namely the sample array, into leaf regions, but using the intermediate regions also to share coding parameters among the corresponding collocated leaf blocks. By this measure, coding procedures performed in tiles - leaf regions - locally, may be associated with coding parameters individually without having to, however, explicitly transmit the whole coding parameters for each leaf region separately. Rather, similarities may effectively exploited by using the multitree subdivision.