VVC Transform Coefficient Coding with Adaptive CBF Context Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional context selection for Context-adaptive binary arithmetic coding (CABAC) of Coded Block Flags (CBFs) in Versatile Video Coding (VVC) is inefficient, as it does not account for the dependency of CBF value probability distribution on transform depth and block type, leading to increased implementation cost and potential degradation in probability adaptation.
Innovation Solution
Implementing improved context models for coding CBFs in VVC, including utilizing specific context models for luma, Cb, and Cr transform units, and selecting context modes based on the value of the Cb transform unit for the current coding unit, to enhance coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional context selection for CABAC of CBFs is used, then the implementation is simpler, but coding efficiency is reduced and probability adaptation is degraded
Solution Approach 1:
The patent applies different context models to different chroma components (Cb and Cr) and different block types (luma, chroma, inter, intra). Specifically, it uses first context models for luma CBFs, second context models for Cb CBFs, and third context models for Cr CBFs, where the third context models are selected based on the values of Cb transform units. This localized differentiation improves probability adaptation and coding efficiency without requiring a single complex universal model.
Solution Approach 2:
The patent introduces dynamic context model selection for Cr CBFs based on the values of Cb transform units in the current coding unit. The context model index for Cr is determined by examining whether Cb transform units contain non-zero coefficients, allowing the context model to adapt dynamically to the actual content being encoded. This dynamic adaptation improves probability estimation accuracy.
2Reliability
If conventional context selection is used, then the system is easier to implement, but probability adaptation is degraded
Solution Approach 1:
The patent implements separate context models for different components and block types to improve probability adaptation. First context models are used for luma CBFs, second for Cb CBFs, and third for Cr CBFs with selection based on Cb transform unit values. This localized approach ensures each component is modeled according to its specific statistical characteristics, improving reliability of probability adaptation.
Solution Approach 2:
The patent uses feedback from Cb transform unit values to select context models for Cr CBFs. The encoder examines the Cb transform unit coefficients and uses this information to determine the appropriate context model index for Cr encoding. This feedback mechanism allows the system to adapt to actual content characteristics, improving probability adaptation reliability.
3Productivity
If conventional context selection is used, then implementation cost is lower, but coding performance is reduced
Solution Approach 1:
The patent applies context models locally to specific components and block types rather than using a single universal model. First context models for luma, second for Cb, and third for Cr with dynamic selection based on Cb values. This localized approach improves coding performance by matching context models to actual data characteristics while keeping implementation costs manageable through structured organization.
Solution Approach 2:
The patent implements dynamic context model selection that adapts to content characteristics. The context model for Cr CBFs is selected based on whether Cb transform units contain non-zero coefficients, allowing the system to use simpler models when appropriate and more adaptive models when needed, optimizing the balance between performance and implementation cost.
Data Source
AI summary
Methods and devices for video coding are provided. In the method, the device determines, for a current luma transform block in a coding unit, whether an intra subpartition (ISP) split type is an ISP mode. In response to determining that the intra subpartition split type is not the ISP mode, the device determines a first context index for a Coded Block Flag (CBF) flag of the current luma transform block as being equal to zero or one, where the CBF flag specifies whether the current luma transform block comprises one or more transform coefficient levels not equal to zero. In response to determining that the intra subpartition split type is the ISP mode, the device determines the first context index based on whether a current transform unit is a first one to be parsed in the coding unit. The device decodes the CBF flag based on the first context index.