Adaptive Binarization in Arithmetic Coding for Changing Symbol Probabilities

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image and video coding standards, such as H.265/HEVC and VP8/VP9, have fixed binarization schemes that do not adapt to changing probability distributions, leading to suboptimal compression efficiency due to the use of pre-defined binarizers tied to context modeling, which can result in longer binary sequences and reduced compression performance.

Innovation Solution

Implementing adaptive binarizer selection methods that allow for the use of modified binarizers based on estimated or historical probability distributions, enabling the selection or construction of binarizers that better match the actual coding distributions, thereby reducing the length of binary sequences and improving compression efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If a fixed pre-defined binarizer is used as prescribed by the standard, then the context modeling is simplified and the implementation is easier, but the binary sequence length increases and compression efficiency deteriorates

Engineering Contradiction:
Improveease of implementationVSAvoidcompression efficiency
Core Design Contradiction:
Ease of manufactureVSLoss of information

Solution Approach 1:

The patent applies dynamics by transitioning from a static, fixed binarizer prescribed by standards to a dynamic, adaptive binarizer that adjusts its parameters based on the actual probability distribution of the data being encoded. The binarizer is updated during the encoding process to match the statistical characteristics of the input data, thereby optimizing compression efficiency while maintaining implementation feasibility through controlled adaptation mechanisms.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent implements parameter changes by modifying the binarizer's internal parameters (such as threshold values and probability estimates) based on the observed data distribution. Instead of using fixed parameters defined by the standard, the system dynamically adjusts these parameters to reflect the actual statistical properties of the encoded data, achieving better compression ratios without fundamentally changing the overall encoding architecture.

Inventive Principle:
Principle #35Parameter changes

2Loss of information

If a modified binarizer is selected based on probability distribution estimation, then the binary sequence length is reduced and compression efficiency is improved, but the device complexity and computational overhead increase

Engineering Contradiction:
Improvecompression efficiencyVSAvoiddevice complexity
Core Design Contradiction:
Loss of informationVSDevice complexity

Solution Approach 1:

The patent applies self-service by enabling the binarizer to automatically adapt to the data distribution without requiring external intervention or complex configuration. The system performs its own probability distribution estimation and adjusts its binarization parameters autonomously during the encoding process, reducing the need for additional complex control mechanisms while achieving improved compression efficiency.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent implements feedback by using the observed frequency of symbols during encoding to continuously update the binarizer's probability estimates. This feedback loop allows the system to refine its binarization strategy based on actual data characteristics, improving compression efficiency while keeping the complexity manageable through iterative adaptation rather than exhaustive optimization.

Inventive Principle:
Principle #23Feedback

3Loss of information

If the binarizer is updated adaptively during encoding, then the compression performance is optimized, but the encoding time and processing speed increase

Engineering Contradiction:
Improvecompression performanceVSAvoidencoding speed
Core Design Contradiction:
Loss of informationVSProductivity

Solution Approach 1:

The patent applies partial action by updating only the necessary parameters of the binarizer based on the current data distribution rather than completely re-optimizing the entire binarization scheme. This selective adaptation approach allows the system to improve compression performance by focusing computational resources on the most impactful parameter adjustments, thereby minimizing the impact on encoding speed.

Inventive Principle:
Principle #16Partial or excessive action

Solution Approach 2:

The patent implements periodic action by updating the binarizer parameters at regular intervals or based on predefined triggers (such as after processing a certain number of symbols or when significant distribution changes are detected). This periodic update strategy balances the need for optimization with the requirement for maintaining encoding speed, avoiding continuous parameter adjustments that would excessively slow down the encoding process.

Inventive Principle:
Principle #19Periodic action

Data Source

PatentEP3182710B1Adaptive binarizer selection for image and video coding
Publication Date: 2021.06.02 BLACKBERRY LTD
  • EP3182710B1 patent drawingFigure 1A~1B
  • EP3182710B1 patent drawingFigure 2A~2B
  • EP3182710B1 patent drawingFigure 3A~3B

AI summary

Methods and devices for image and video coding using arithmetic coding. The binarization of symbols in an encoder and decoder is adaptive based on changes to the probability distribution as symbols are encoded/decoded. A binarizer may be generated based upon a probability distribution, used to binarize a symbol, and then the probability distribution is updated based on the symbol. Updates to the binarizer may be made after each symbol, after a threshold number of symbols, or once the updated probability distribution differs by more than a threshold amount from the probability distribution used in generating the current binarizer. The probability distributions may be context-specific.