Arithmetic Coding Context Mapping for Spectral Coefficient Latency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The high number of contexts and corresponding probability density functions required for arithmetic coding leads to increased encoding/decoding latency and memory capacity requirements, necessitating a solution that maintains compression efficiency while reducing these constraints.
Innovation Solution
The method involves using context classes instead of individual contexts to determine probability density functions by non-uniformly quantizing preceding spectral coefficients, grouping similar contexts into a single class and mapping them to a single probability density function, and employing a two-step non-linear quantization process to adapt to variance, with context class determination based on the output of the second quantization step.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If context based arithmetic coding with large neighbourhood is used to improve compression, then compression efficiency is improved, but the number of contexts and probability density functions increases dramatically
Solution Approach 1:
The patent merges multiple similar contexts into a single context class. Instead of maintaining separate probability density functions for each individual context, the invention groups contexts with similar characteristics together and uses a single representative probability density function for the entire group, thereby reducing the total number of probability density functions required
Solution Approach 2:
The patent changes the parameter representation by using non-uniform quantization of spectral coefficients to define context classes. This transformation converts the continuous spectral coefficient values into discrete quantized levels, which then serve as indices for selecting context classes and their associated probability density functions
2Loss of information
If the number of probability density functions is increased to handle more contexts, then compression efficiency is improved, but memory capacity requirements increase
Solution Approach 1:
Multiple contexts that would otherwise require separate probability density functions are merged into context classes. Each context class stores only one representative probability density function, significantly reducing the memory capacity required to store all probability density functions while maintaining adequate compression performance
3Loss of information
If more contexts are handled with higher precision, then compression efficiency is improved, but encoding/decoding latency increases
Solution Approach 1:
The patent applies non-uniform quantization to transform spectral coefficient values into discrete quantized levels. This parameter transformation simplifies the context identification process by converting continuous values into discrete indices, enabling faster context class determination and probability density function selection, thereby reducing encoding/decoding latency
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention proposes a method and a device for arithmetic encoding and decoding of a current spectral coefficient using preceding spectral coefficients. Said preceding spectral coefficients are already encoded and both, said preceding and current spectral coefficients, are comprised in one or more quantized spectra resulting from quantizing time-frequency-transform of video, audio or speech signal sample values. Said method comprises processing the preceding spectral coefficients, using the processed preceding spectral coefficients for determining a context class being one of at least two different context classes, using the determined context class and a mapping from the at least two different context classes to at least two different probability density functions for determining the probability density function, and arithmetic encoding the current spectral coefficient based on the determined probability density function wherein processing the preceding spectral coefficients comprises non-uniformly quantizing absolutes of the preceding spectral coefficients for use in determining of the context class.