Context Set Selection for Multi-Level Video Significance Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video encoding standards, such as H.264/AVC and developing MPEG-H, face inefficiencies in encoding and decoding significance maps due to the high percentage of data required for encoding quantized transform coefficients, particularly in larger transform units, leading to suboptimal compression efficiency.
Innovation Solution
Implementing multi-level significance maps that partition transform units into regions based on coefficient groups, using context-adaptive encoding and decoding to determine context sets for significant-coefficient flags, reducing the need for explicit encoding of certain flags and improving entropy encoding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional context models are used for encoding significance maps in larger transform units, then the encoding process can be implemented with existing standards, but the compression efficiency deteriorates due to high data requirements
Solution Approach 1:
The transform unit is divided into multiple coefficient groups, and the significance map is segmented into multiple regions based on these coefficient groups. Each region is encoded using a separate context model, allowing more efficient compression by processing smaller, more manageable segments rather than the entire transform unit at once.
Solution Approach 2:
Different context models are assigned to different regions of the significance map based on local characteristics. The context model selection is adapted to the specific region being encoded, allowing optimization of compression efficiency for each local area rather than using a single uniform context model for the entire transform unit.
2Measurement precision
If separate context models are assigned to each coefficient position, then encoding precision can be maintained, but device complexity increases due to tracking and looking up large numbers of different contexts
Solution Approach 1:
The significance map is divided into multiple regions, and context models are segmented accordingly. Instead of managing a single large context model for all coefficient positions, the system uses multiple smaller context models associated with different regions, reducing the complexity of context tracking and lookup operations.
Solution Approach 2:
Each region is assigned a context model that is optimized for its local characteristics. This allows the system to maintain encoding precision for each region while using simpler, more manageable context models rather than a single complex context model that must accommodate all possible coefficient positions.
3Productivity
If multi-level significance maps are implemented, then compression efficiency improves for larger transform units, but the encoding and decoding processes become more complex
Solution Approach 1:
The significance map is divided into multiple levels or regions, with each level processed using its own context model. This segmentation allows the system to achieve better compression efficiency for larger transform units by handling them as multiple smaller segments, while the complexity is managed through a systematic approach to processing each segment.
Solution Approach 2:
The context model selection is made dynamic, adapting to the specific region being encoded. The system can switch between different context models based on the current processing stage and region, allowing flexible optimization of compression efficiency while managing complexity through adaptive rather than static processing.
Data Source
AI summary
Methods of encoding and decoding for video data are described for encoding or decoding multi-level significance maps. Distinct context sets may be used for encoding the significant-coefficient flags in different regions of the transform unit. In a fixed case, the regions are defined by coefficient group borders. In one example, the upper-left coefficient group is a first region and the other coefficient groups are a second region. In a dynamic case, the regions are defined by coefficient group borders, but the encoder and decoder dynamically determine in which region each coefficient group belongs. Coefficient groups may be assigned to one region or another based on, for example, whether their respective significant-coefficient-group flags were inferred or not.