Video Decoder EOB Position Embedding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current video encoding techniques require significant computational resources and bandwidth for decoding transform blocks of quantized coefficients, as they involve extensive context-coded binary decisions, particularly when dealing with large transform blocks.

Innovation Solution

The method involves partitioning quantized transform coefficients into two groups and embedding information about end-of-block (EOB) positions within the coefficients, allowing for reduced decoding operations by skipping the encoding and decoding of EOB tokens for certain coefficients based on the sum and parity of the first coefficients group, thereby optimizing the coefficient token tree traversal.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If conventional decoding methods are used for transform blocks, then complete accuracy is maintained, but decoding complexity and computational resources increase significantly

Engineering Contradiction:
Improvedecoding complexityVSAvoidcoefficient decoding accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The transform block is divided into multiple coefficient groups, with the first group decoded using full precision and subsequent groups decoded using simplified methods. This segmentation allows the patent to reduce decoding complexity for less critical coefficients while maintaining accuracy for important ones, directly resolving the contradiction between complexity and precision.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Different decoding strategies are applied to different regions of the coefficient block. The first coefficient group receives full decoding treatment while subsequent groups use simplified decoding with EOB position embedding. This local differentiation optimizes the balance between accuracy and complexity based on the importance of different coefficient regions.

Inventive Principle:
Principle #3Local quality

2Loss of information

If EOB tokens are encoded for all coefficients, then complete information is preserved, but the number of bits required increases

Engineering Contradiction:
Improvecoefficient information completenessVSAvoidnumber of bits
Core Design Contradiction:
Loss of informationVSQuantity of substance

Solution Approach 1:

The patent extracts and embeds EOB position information within the coefficient values themselves rather than encoding separate EOB tokens for each coefficient. This extraction of EOB information from the token stream and embedding it in the coefficient data reduces the total number of bits required while preserving the necessary end-of-block information.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The EOB position information is merged with the coefficient data by embedding it within the coefficient values. Instead of separate EOB tokens, the end-of-block positioning is combined with the coefficient representation, reducing redundancy and decreasing the total bit quantity while maintaining information completeness.

Inventive Principle:
Principle #5Merging (Combining)

3Reliability

If context-coded binary decisions are performed for each coefficient, then decoding accuracy is maintained, but computational resources and bandwidth consume more

Engineering Contradiction:
Improvedecoding reliabilityVSAvoidcomputational resource consumption
Core Design Contradiction:
ReliabilityVSUse of energy by moving object

Solution Approach 1:

The coefficient decoding process is segmented into two stages: full context-coded binary decision decoding for the first group, and simplified decoding with embedded EOB positions for subsequent groups. This segmentation reduces the number of context-coded decisions required, lowering computational resource consumption while maintaining reliability for the most important coefficients.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies full context-coded binary decision decoding only partially - specifically for the first coefficient group - rather than for all coefficients. This partial application of the computationally intensive decoding method maintains reliability where needed while reducing overall computational resource consumption and bandwidth usage.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP3989571A1Embedding information about EOB positions
Publication Date: 2022.04.27 GOOGLE LLC
  • EP3989571A1 patent drawingFigure 1
  • EP3989571A1 patent drawingFigure 2
  • EP3989571A1 patent drawingFigure 3

AI summary

A method for decoding a transform block of quantized transform coefficients includes: decoding, from an encoded bitstream, a predetermined number of coefficients of the quantized transform coefficients, wherein the transform block comprises the predetermined number of coefficients and subsequent quantized transform coefficients; determining a value for the predetermined number of coefficients; decoding, from the encoded bitstream, a subsequent quantized transform coefficient of the subsequent quantized transform coefficients; and determining whether to decode an end-of-block (EOB) indicator based on the value that is determined for the predetermined number of coefficients.