Spectral Coefficient Arithmetic Coding with Quantized Context Classes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Arithmetic coding for multimedia data compression faces challenges with high encoding/decoding latency and memory capacity requirements due to the large number of contexts and corresponding probability density functions that need to be handled.
Innovation Solution
The method involves using preceding spectral coefficients to determine context classes, which are then mapped to probability density functions for encoding/decoding, employing non-uniform quantization and variance determination to reduce the complexity and number of contexts, and utilizing a look-up table or hash table for efficient mapping.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a large neighbourhood of preceding coefficients is used for context-based arithmetic coding, then compression efficiency is improved, but the number of different possible contexts explodes combinatorially, leading to increased memory capacity requirements and encoding/decoding latency
Solution Approach 1:
The patent merges multiple similar contexts into a single context class by grouping contexts that share the same non-uniformly quantized absolute values of preceding coefficients. This consolidation reduces the combinatorial explosion of contexts while preserving the essential statistical characteristics needed for effective arithmetic coding, thereby maintaining compression efficiency with reduced memory and computational requirements.
Solution Approach 2:
The patent applies non-uniform quantization to the absolute values of preceding coefficients, transforming the continuous or high-resolution coefficient values into a reduced set of discrete quantization levels. This parameter transformation dramatically reduces the number of distinct context patterns while retaining the most significant variations, enabling efficient context classification without sacrificing compression performance.
2Manufacturing precision
If more probability density functions are stored and handled to cover all possible contexts, then coding accuracy is improved, but encoding/decoding latency and memory capacity requirements increase
Solution Approach 1:
The patent consolidates multiple probability density functions into a smaller set by grouping contexts into classes. Each context class is associated with a single representative probability density function that models the statistical characteristics of all contexts in that class. This merging reduces the number of PDFs that need to be stored and retrieved, thereby decreasing encoding/decoding latency while maintaining sufficient coding accuracy through the representative nature of each PDF.
Solution Approach 2:
The patent transforms the high-dimensional context space into a lower-dimensional classification space using non-uniform quantization. This parameter transformation enables the use of a reduced number of probability density functions by ensuring that contexts with similar quantized values share similar statistical properties, thus requiring fewer distinct PDFs to achieve accurate coding.
3Productivity
If a large number of different contexts are handled with correspondingly complex mapping, then compression performance is improved, but memory capacity requirements and device complexity increase
Solution Approach 1:
The patent applies non-uniform quantization to transform the preceding coefficient values into a reduced set of discrete levels. This parameter transformation compresses the information content of the context description, reducing the memory required to store and process context information while preserving the essential statistical patterns needed for high compression performance.
Solution Approach 2:
The patent merges multiple fine-grained contexts into coarser context classes based on shared quantized characteristics. This merging reduces the total number of context entries that must be stored in memory, decreasing memory capacity requirements while maintaining compression performance through the aggregated statistical representation of each context class.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention proposes a method and a device for arithmetic encoding 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.