Adaptive Binarizer Selection for Shorter Arithmetic Coding Bins

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 default binarizers that do not account for actual data distributions.

Innovation Solution

The method involves selecting or constructing a modified binarizer based on the actual distribution of data, signaled through header information, to reduce the binary sequence length and improve compression efficiency by dynamically adjusting the binarization process.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If a fixed binarization scheme is used as prescribed by the standard, then the decoding logic remains simple and consistent, but the compression efficiency deteriorates due to inability to adapt to changing probability distributions

Engineering Contradiction:
Improvecompression efficiencyVSAvoidbinarizer selection complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent implements dynamic binarizer selection by allowing the encoder to choose from multiple predefined binarizers (first, second, third binarizers) based on the actual probability distribution of the data being encoded. The decoder mirrors this selection capability, enabling adaptive binarization that changes with data characteristics while maintaining synchronized operation between encoder and decoder.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent changes the parameter of binarization scheme selection by introducing multiple binarizer options with different structures (different numbers of bins for various symbol ranges). The system selects which binarizer to use based on data characteristics, effectively changing the binarization parameters to match the probability distribution and improve compression efficiency.

Inventive Principle:
Principle #35Parameter changes

2Productivity

If a default binarizer is used for all data, then the implementation remains simple and standardized, but the binary sequence length increases reducing compression efficiency

Engineering Contradiction:
Improvebinary sequence length reductionVSAvoidadaptability to data distribution
Core Design Contradiction:
ProductivityVSAdaptability or versatility

Solution Approach 1:

The patent applies local quality by having different binarizers optimized for different data characteristics. Each binarizer (first, second, third) has a specific structure suited for certain probability distributions. The system selects the appropriate binarizer locally based on the actual data being encoded, rather than applying a uniform binarization scheme globally.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The system dynamically selects which binarizer to use based on the probability distribution of the input data. This dynamic adaptation allows the binarization process to optimize for the specific data characteristics encountered, reducing binary sequence length when the data distribution matches the optimized binarizer structure.

Inventive Principle:
Principle #15Dynamics

3Productivity

If multiple binarizers are provided for selection, then compression efficiency improves through adaptive binarization, but the device complexity and implementation difficulty increase

Engineering Contradiction:
Improvecompression efficiencyVSAvoidimplementation ease
Core Design Contradiction:
ProductivityVSEase of manufacture

Solution Approach 1:

The patent implements universality by providing multiple binarizers that can handle different data scenarios within a single encoding/decoding system. Each binarizer serves a specific function optimized for certain probability distributions, but collectively they provide a universal solution that adapts to various data types and characteristics without requiring separate systems.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent applies preliminary action by pre-defining multiple binarizer structures in advance, each optimized for specific data characteristics. Rather than dynamically creating binarizers during operation, the system prepares multiple binarization schemes beforehand and selects from them, reducing implementation complexity while maintaining adaptability.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3182705B1Binarizer selection for image and video coding
Publication Date: 2021.11.17 BLACKBERRY LTD
  • EP3182705B1 patent drawingFigure 1A~1B
  • EP3182705B1 patent drawingFigure 2A~2B
  • EP3182705B1 patent drawingFigure 3A~3B

AI summary

Methods and devices for image and video coding using arithmetic coding. The binarization of symbols, normally carried out using a default binarizer, may be carried out using a modified binarizer. The binarizer may be selected or constructed at the encoder and/or decoder to improve the efficiency of binarization and minimize the length of the sequence of bins output by the binarizer for an input sequence of symbols. The modified binarizer may be selected from predefined binarizers. The modified binarizer may be constructed through modifying the assignment of symbols to leaf nodes of a default coding tree used by the default binarizer. The modified binarizer may be constructed by modifying the structure of a default coding tree used by the default binarizer. Changes to the default coding tree may be signaled to the decoder in header information. Alternatively, the decoder may construct the modified binarizer based on probability information sent by the encoder in a header.