Transform Coefficient Correction Bounds for Quantization-Constrained Filtering

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing video coding technologies face issues with sub-optimal Mean Squared Error (MSE) due to the loss of information during in-loop filtering processes, as conventional quantization constraints are not satisfied after filtering, leading to sub-optimal reconstruction of video data.

Innovation Solution

A method and apparatus for correcting transform coefficients using integer-based computation of inferior and superior correction bounds, ensuring the quantization constraint is satisfied by applying a correction function to the coefficients, thereby maintaining the integrity of the quantization levels.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If in-loop filtering is applied to improve visual quality, then filtering effectiveness is improved, but quantization constraint satisfaction deteriorates leading to sub-optimal MSE

Engineering Contradiction:

Improvefiltering effectivenessVSAvoidquantization constraint satisfaction

Core Design Contradiction:

ReliabilityVSManufacturing precision

Solution Approach 1:

The patent implements a feedback mechanism where the quantization constraint is checked after filtering, and a correction is applied if the constraint is violated. The correction amount is determined by comparing the filtered coefficient against the quantization bound, creating a closed-loop system that ensures constraint satisfaction while preserving filtering benefits

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The patent modifies the filtered transform coefficient by adding a correction term that changes its parameter values. The correction is calculated as the difference between the filtered coefficient and the quantization bound, effectively adjusting the coefficient to satisfy the quantization constraint while maintaining the filtering effect

Inventive Principle:

Principle #35Parameter changes

2Ease of manufacture

If conventional quantization constraints are applied, then encoding simplicity is maintained, but reconstruction fidelity deteriorates due to information loss during filtering

Engineering Contradiction:

Improveencoding simplicityVSAvoidreconstruction fidelity

Core Design Contradiction:

Ease of manufactureVSLoss of information

Solution Approach 1:

The patent performs a preliminary correction of the transform coefficient before final quantization encoding. By anticipating the quantization constraint violation that would occur after filtering, the method pre-adjusts the coefficient to ensure constraint satisfaction, thereby preventing information loss rather than correcting it afterward

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent introduces a correction term as an intermediary between the filtered coefficient and the final quantized value. This intermediary element bridges the gap between the filtered signal and the quantization constraint, allowing both filtering effectiveness and constraint satisfaction to coexist without direct conflict

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentEP4679831A1Integer computation of correction bounds for quantization constrained correction

Publication Date: 2026.01.14 INTERDIGITAL CE PATENT HOLDINGS SAS

Figure 1
Figure 2
Figure 3

AI summary

A method, performed by a video encoder or a video decoder, for correcting a transform coefficient associated with a quantization level is disclosed. An inferior correction bound and a superior correction bound are computed from quantization step parameters, the quantization level, minimum and maximum bound offsets each equal to a ratio between an integer offset parameter and an integer power of two. A correction function is then applied to the transform coefficient based on the computed inferior and superior correction bounds.