Lossless Coding Block Size Alignment in Video Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding standards, such as VVC, face inefficiencies in lossless coding modes, including unsupported residual block sizes greater than 32x32, suboptimal residual coding schemes, and inefficient context modeling for transform coefficient coding.
Innovation Solution
Proposed methods include aligning the maximum residual coding block size for lossless CUs with the maximum block size supported by transform skip mode, selecting residual coding schemes adaptively based on conditions, disabling complex decoder-side tools like DMVR and BDOF in lossless coding, and optimizing context modeling and sign flag coding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the maximum residual coding block size for lossless CUs is increased beyond 32x32, then the coding efficiency is improved, but the compatibility with transform skip mode is lost
Solution Approach 1:
The patent introduces dynamic block size support for residual coding in lossless mode by allowing the residual transform block size to differ from the CU size. The transform unit (TU) can be larger than 32x32 while the CU size remains flexible, enabling adaptive block sizing that maintains compatibility with transform skip mode while improving coding efficiency for larger blocks.
Solution Approach 2:
The patent separates the CU partitioning from the residual transform block sizing. The CU is divided into PUs and TUs independently, allowing the residual transform block to have a size that is not constrained by the CU size. This segmentation enables the residual coding to operate on optimally-sized blocks regardless of the CU dimensions.
2Manufacturing precision
If complex decoder-side tools like DMVR and BDOF are enabled in lossless coding, then the video quality is improved, but the decoding complexity and computational load increase
Solution Approach 1:
The patent applies decoder-side tools selectively based on local characteristics. The tools DMVR and BDOF are enabled only when specific conditions are met, such as when certain prediction modes are used or when the block size and content characteristics warrant their application. This selective application maintains video quality where needed while reducing overall decoding complexity.
3Ease of manufacture
If the residual coding scheme is fixed for lossless mode, then the implementation is simplified, but the coding efficiency for different block sizes and content types deteriorates
Solution Approach 1:
The patent introduces dynamic selection of residual coding schemes based on block size, content characteristics, and coding conditions. The encoder can adaptively choose between different residual coding methods (e.g., transform coding, skip-mode residual coding, or other schemes) to optimize performance for each specific block, rather than using a fixed scheme for all lossless mode blocks.
4Productivity
If context modeling is optimized for transform coefficient coding, then the coding efficiency is improved, but the model complexity and signaling overhead increase
Solution Approach 1:
The patent optimizes context modeling by adapting the context models based on the transform type and block characteristics. Different context models are used for different transform types (e.g., DCT, DST), and the context model parameters are adjusted according to the block size and position. This parameter adaptation improves coding efficiency without requiring extensive additional signaling, as the adjustments are based on already-available block metadata.
Data Source
AI summary
Methods, apparatuses, and non-transitory computer-readable storage mediums are provided for decoding a video signal. A decoder partitions a video picture into a plurality of coding units (CUs) comprising a lossless CU. The decoder may further receive a high-level syntax. The high-level syntax may include, for example, a first flag that indicates whether a residual coding scheme is switchable.