Adaptive Probability Estimation Mixture for Video Entropy Coding

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing video compression techniques face challenges in accurately estimating cumulative probability distributions for multi-symbol arithmetic coding, leading to suboptimal coding efficiency due to either fast adaptation's high variance or slow adaptation's inability to react to changes in probability statistics.

Innovation Solution

The use of a mixture of two or more adaptive cumulative probability estimates with different update parameters allows for a balance between fast and slow adaptation, incorporating a weighted sum of estimates to quickly adapt to changes while maintaining stability, and includes mechanisms to reset or adjust adaptation behavior based on divergence or variance signals.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If fast adaptation is used for probability estimation, then the ability to react to changes in probability statistics is improved, but the variance of the estimate increases leading to suboptimal coding efficiency

Engineering Contradiction:

Improveability to react to changes in probability statisticsVSAvoidvariance of probability estimate

Core Design Contradiction:

Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The patent combines multiple adaptive probability estimates with different update parameters (α values) into a single mixed estimate. This merging allows the system to benefit from both fast adaptation (when probability changes occur) and slow adaptation (when stability is needed), resolving the contradiction between adaptability and precision.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The system dynamically adjusts the mixing parameter λ based on the observed variance and divergence of individual estimates. When variance is high, the system increases reliance on more stable estimates; when variance is low, it allows faster adaptation. This dynamic adjustment resolves the contradiction by making the adaptation speed conditional rather than fixed.

Inventive Principle:

Principle #15Dynamics

2Stability of the object's composition

If slow adaptation is used for probability estimation, then the stability of the estimate is improved, but the ability to react to changes in probability statistics deteriorates

Engineering Contradiction:

Improvestability of probability estimateVSAvoidability to react to changes in probability statistics

Core Design Contradiction:

Stability of the object's compositionVSAdaptability or versatility

Solution Approach 1:

The patent merges multiple estimates with different adaptation speeds, allowing the system to maintain stability from slow adaptation while capturing changes through fast adaptation components. The mixed estimate benefits from both characteristics simultaneously.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The system uses feedback from the variance and divergence of individual estimates to adjust the mixing parameter λ. When changes in probability statistics are detected through increased variance, the feedback mechanism increases the weight of faster-adapting estimates, enabling the system to react to changes while maintaining overall stability.

Inventive Principle:

Principle #23Feedback

3Productivity

If multiple adaptive probability estimates are combined, then coding efficiency is improved, but the complexity of the implementation increases

Engineering Contradiction:

Improvecoding efficiencyVSAvoidcomplexity of probability estimation

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The patent manages complexity by parameterizing the mixing process through a single control parameter λ that can be adjusted based on observed variance. This allows multiple estimates to be combined efficiently without requiring complex decision logic, maintaining implementation simplicity while achieving improved coding efficiency.

Inventive Principle:

Principle #35Parameter changes

4Productivity

If adaptive probability estimation is used, then coding efficiency is improved, but the computational resources and thermal performance are affected

Engineering Contradiction:

Improvecoding efficiencyVSAvoidcomputational resources and thermal performance

Core Design Contradiction:

ProductivityVSUse of energy by moving object

Solution Approach 1:

The system performs partial adaptation by combining estimates rather than fully adapting a single complex model. This partial action approach achieves most of the coding efficiency benefits while reducing the computational burden and thermal impact of full adaptation of a single high-complexity model.

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS20240129473A1Probability estimation in multi-symbol entropy coding

Publication Date: 2024.04.18 INTEL CORP

AI summary

Entropy coding, such as multi-symbol arithmetic coding, is used in video compression to encode data into a compressed bit stream for transmission. Some entropy coding techniques are adaptive, meaning that the probability distribution is updated on the fly, based on the data. Accuracy of cumulative probability estimation in adaptive multi-symbol arithmetic coding can impact coding efficiency. To address the issue, a mixture of two or more adaptive cumulative probability estimations computed using two or more adaptation parameters can be used in place of a single cumulative probability estimate. The two or more adaptation parameters can be unique for a context model. A divergence in the adaptive cumulative probability estimations may signal a sudden change in the probability of a symbol. The divergence may trigger a reset of one or more adaptive cumulative probability estimations.