Syntax-Enabled Neural Post Filter for Energy-Aware Video Recovery

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The increasing demand for high-resolution and high-dynamic-range displays has led to a corresponding increase in energy consumption, making it necessary to develop energy-aware images or videos that require less energy when displayed, while maintaining image quality.

Innovation Solution

A method and apparatus that utilize an energy recovery neural network post-processing filter, enabled by a flag in the supplemental enhancement information message, to decode and process video data, reducing energy consumption and allowing receivers to recover the original image from an energy-aware image.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If high-resolution and high-dynamic-range displays are used, then image quality is improved, but energy consumption increases

Engineering Contradiction:
Improveimage qualityVSAvoidenergy consumption
Core Design Contradiction:
Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent transforms video content into an energy-aware representation by modifying pixel value parameters. The energy-aware video encoder adjusts the luminance values of pixels to reduce overall energy consumption while maintaining visual quality through perceptual optimization, directly addressing the contradiction between image quality and energy consumption.

Inventive Principle:
Principle #35Parameter changes

2Use of energy by moving object

If energy-aware video encoding is applied, then energy consumption is reduced, but image quality deteriorates

Engineering Contradiction:
Improveenergy consumptionVSAvoidimage quality
Core Design Contradiction:
Use of energy by moving objectVSMeasurement precision

Solution Approach 1:

The system incorporates a feedback mechanism where the energy-aware video decoder reconstructs the original video content from the energy-aware representation and compares it with the source. This feedback loop enables optimization of the encoding process to minimize quality loss while achieving energy reduction goals, resolving the contradiction between energy efficiency and image quality.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The energy-aware video encoder performs preliminary transformation of the video content into an energy-optimized representation before transmission or storage. By pre-processing the video to reduce energy consumption while preserving essential visual information, the system prepares the content for efficient display without requiring real-time processing during playback.

Inventive Principle:
Principle #10Preliminary action

3Loss of information

If neural network post-processing filter is applied, then original image can be recovered, but processing complexity increases

Engineering Contradiction:
Improveimage recoveryVSAvoidprocessing complexity
Core Design Contradiction:
Loss of informationVSDevice complexity

Solution Approach 1:

The patent replaces traditional complex image recovery algorithms with a neural network-based post-processing filter. The neural network, trained to reconstruct original video content from energy-aware representations, provides efficient and accurate image recovery with reduced computational complexity compared to conventional methods, resolving the contradiction between information recovery and processing complexity.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentEP4633151A1Neural network post filter for energy-recovery
Publication Date: 2025.10.15 INTERDIGITAL CE PATENT HOLDINGS SAS
  • EP4633151A1 patent drawingFigure 1~2
  • EP4633151A1 patent drawingFigure 3
  • EP4633151A1 patent drawingFigure 4

AI summary

Methods and devices enable the distribution of video with reduced energy and the reconstruction of a video with recovered energy, similar to the original video. The recovery is done using a neural network post filter enabled through a syntax element carried by the video. An additional syntax element may define a recovery ratio.