Multi-Level Significance Maps for Large Transform Unit Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video encoding standards, such as H.264/AVC and HEVC, face inefficiencies in encoding and decoding significance maps due to the computational intensity and memory bandwidth requirements for larger transform units like 16×16 and 32×32, particularly in determining context for significant-coefficient flags, which affects the overall encoding efficiency.
Innovation Solution
The use of multi-level significance maps, where significant-coefficient flags are grouped into larger transform units (e.g., 16×16 and 32×32) and encoded using higher level significance maps (L1 maps) to reduce the number of contexts needed, thereby reducing computational load and memory access costs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional single-level significance maps are used for encoding, then the encoding process can be implemented with existing standards, but the computational complexity and memory bandwidth requirements increase significantly for larger transform units (16×16 and 32×32)
Solution Approach 1:
The patent divides the significance map into multiple levels by grouping coefficient positions into different hierarchical levels. Level 0 contains individual coefficient positions, while level 1 contains groups of level 0 positions. This segmentation allows the encoder to process smaller units (level 0) separately from larger units (level 1), reducing the computational burden on any single processing stage while maintaining support for various transform unit sizes.
Solution Approach 2:
The patent introduces a new hierarchical dimension to the traditional flat significance map structure. By adding level 1 significance map that groups and summarizes level 0 coefficient positions, the system transforms the two-dimensional coefficient matrix into a three-dimensional hierarchical structure. This dimensional change enables more efficient encoding by allowing skip operations and reducing the effective search space for context modeling.
2Measurement precision
If context determination is performed for each coefficient position in traditional significance maps, then encoding accuracy is maintained, but memory access costs and processing time increase
Solution Approach 1:
The patent applies partial action by not requiring full context determination for all coefficient positions. Instead, the level 1 significance map allows the encoder to skip context determination for entire groups of coefficients when certain conditions are met (e.g., when all coefficients in a group are zero or when probability models indicate low likelihood of significance). This selective approach maintains encoding accuracy for important coefficients while significantly reducing processing time for less significant regions.
Solution Approach 2:
The level 1 significance map performs preliminary grouping and summarization of coefficient positions before the detailed level 0 encoding begins. By pre-organizing coefficients into hierarchical groups and determining level 1 significance flags in advance, the system prepares the encoding structure beforehand, allowing faster processing during the actual level 0 encoding phase without compromising accuracy.
3Adaptability or versatility
If individual coefficient positions are encoded separately in traditional significance maps, then encoding flexibility is maintained, but the number of contexts to track increases substantially
Solution Approach 1:
The patent merges multiple individual coefficient positions into grouped units at level 1. Instead of tracking separate contexts for each coefficient position in the traditional approach, the level 1 significance map combines and summarizes the significance status of multiple level 0 coefficients into single level 1 entries. This merging reduces the total number of contexts that need to be tracked and managed, simplifying the encoder's memory requirements while preserving the flexibility to encode individual coefficients at level 0 when necessary.
Data Source
AI summary
Methods of encoding and decoding for video data are described in which multi-level significance maps are used in the encoding and decoding processes. The significant-coefficient flags that form the significance map are grouped into contiguous groups, and a significant-coefficient-group flag signifies for each group whether that group contains no non-zero significant-coefficient flags. If there are no non-zero significant-coefficient flags in the group, then the significant-coefficient-group flag is set to zero. The set of significant-coefficient-group flags is encoded in the bitstream. Any significant-coefficient flags that fall within a group that has a significant-coefficient-group flag that is non-zero are encoded in the bitstream, whereas significant-coefficient flags that fall within a group that has a significant-coefficient-group flag that is zero are not encoded in the bitstream.