Intra Prediction Mode Derivation Using Content-Based Gradient Analysis

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing video coding standards like VVC and ECM face challenges in accurately deriving Intra Prediction Mode (IPM) and Low Frequency Non-Separable Transform (LFNST) modes, as they rely on split lines that represent object boundaries rather than content representation, leading to inefficiencies in coding efficiency.

Innovation Solution

The proposed method utilizes analysis tools like Directional Intra Mode Derivation (DIMD) and Template-based Intra Mode Derivation (TIMD) to analyze samples within and around prediction blocks, weighting gradients from specific block locations to determine dominant directions for IPM and LFNST modes, enhancing accuracy and coding efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If split lines representing object boundaries are used to derive IPM and LFNST modes in existing video coding standards, then the derivation process is simplified, but coding efficiency deteriorates due to inaccurate content representation

Engineering Contradiction:
Improvesimplicity of mode derivationVSAvoidcoding efficiency
Core Design Contradiction:
Ease of manufactureVSProductivity

Solution Approach 1:

The patent replaces the mechanical/geometric approach of using split lines (object boundaries) with a content-based analysis approach. By substituting the derivation mechanism from boundary-based to content-based gradient analysis, the system achieves more accurate mode derivation that reflects actual image content rather than arbitrary boundary structures.

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

Solution Approach 2:

The patent changes the fundamental parameter used for mode derivation from spatial position (split line locations) to content characteristics (gradient directions and magnitudes). This parameter transformation enables the derivation process to adapt to actual image content variations, improving coding efficiency while maintaining computational feasibility.

Inventive Principle:
Principle #35Parameter changes

2Productivity

If content-based analysis is performed to accurately determine IPM and LFNST modes, then coding efficiency improves, but computational complexity increases

Engineering Contradiction:
Improvecoding efficiencyVSAvoidcomputational complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent segments the prediction block into multiple regions and performs gradient analysis on specific sub-blocks rather than the entire block. This segmentation strategy reduces the total number of gradient calculations required while still capturing essential content characteristics for accurate mode derivation.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs gradient analysis on selected portions of the prediction block rather than exhaustively analyzing all pixels. By applying partial action to representative regions, the system achieves sufficient accuracy for mode derivation with reduced computational burden compared to complete block analysis.

Inventive Principle:
Principle #16Partial or excessive action

3Measurement precision

If gradient analysis is performed on the entire prediction block to determine dominant directions, then measurement precision improves, but processing time increases

Engineering Contradiction:
Improveaccuracy of dominant direction determinationVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent applies different analysis strategies to different regions of the prediction block based on local characteristics. By identifying and analyzing only the most informative local regions for gradient computation, the system maintains high precision in dominant direction determination while avoiding unnecessary computations in less critical areas.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent performs preliminary analysis to identify candidate regions or dominant patterns before conducting full gradient analysis. This preliminary action enables the system to focus subsequent detailed analysis on specific areas of interest, reducing overall processing time while preserving measurement precision where it matters most.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentUS20260019570A1Derivation of intra prediction mode modes using content analysis
Publication Date: 2026.01.15 NOKIA TECHNOLOGIES OY
  • US20260019570A1 patent drawing
  • US20260019570A1 patent drawing
  • US20260019570A1 patent drawing

AI summary

An apparatus includes at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to: analyze an analysis area associated with a prediction block of a coding unit; and derive at least one intra prediction mode, based on the analyzing of the analysis area associated with the prediction block of the coding unit that does not have a directionality from its coding mode.