Video Decoder Normal-Prediction Fallback for Absent Weights

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing video coding and decoding technologies perform weighted prediction processing even when no weight coefficient and offset value are present for reference lists, leading to incorrect prediction processing.

Innovation Solution

Implement a video decoding apparatus with separate weighted and normal prediction processing units, where the normal processing unit generates predictions when weight and offset values are absent for bi-prediction or uni-prediction, ensuring accurate prediction image generation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If weighted prediction processing is always performed when weightedPredFlag is 1, then the prediction processing is simplified, but incorrect prediction results occur when weight coefficients and offset values are absent

Engineering Contradiction:
Improveprediction processing complexityVSAvoidprediction accuracy
Core Design Contradiction:
Device complexityVSReliability

Solution Approach 1:

The patent changes the control parameter logic by introducing a new condition check: when weightedPredFlag is 1 but weight coefficients and offset values are absent for both L0 and L1 lists, the system switches from weighted prediction to normal prediction mode. This parameter change resolves the contradiction by making the prediction mode adaptive based on the actual availability of weighting parameters, ensuring both processing simplicity and prediction accuracy.

Inventive Principle:
Principle #35Parameter changes

2Reliability

If normal prediction processing is performed instead of weighted prediction when weight information is absent, then prediction accuracy is improved, but processing complexity increases due to separate processing units

Engineering Contradiction:
Improveprediction accuracyVSAvoidprocessing unit structure
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent implements dynamic processing by making the prediction mode selectable rather than fixed. The prediction processing unit dynamically switches between weighted prediction and normal prediction based on runtime conditions (availability of weight coefficients and offset values). This dynamic approach avoids the need for completely separate processing units while maintaining both accuracy and simplicity.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentUS12418674B2Video coding apparatus and video decoding apparatus
Publication Date: 2025.09.16 SHARP KK
  • US12418674B2 patent drawing
  • US12418674B2 patent drawing
  • US12418674B2 patent drawing

AI summary

In the related art, in some cases the weighted prediction processing is performed even in a case that normal prediction processing is to be performed. A video decoding apparatus includes a weighted prediction processing unit configured to decode a weight coefficient and an offset value from coded data and to generate a prediction image by multiplying an interpolation image by the weight coefficient and adding the offset value to the interpolation image, and a normal prediction processing unit configured to generate a prediction image from the interpolation image. In a case that in the weighted prediction processing unit, a bi-prediction is performed and information of the weight coefficient and the offset value is absent for both a reference list 0 and a reference list 1, the normal prediction processing unit generates a bi-prediction image.