Sliding Window Context Derivation for Transform Coefficients

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing video coding techniques face inefficiencies in determining context for transform coefficients due to the computational expense of position-based context support neighborhoods, which require determining relative locations and accessing extensive data, especially when implementing parallel processing and handling large blocks or sub-blocks.

Innovation Solution

The approach involves defining a context derivation neighborhood based on the transform coefficient scan order, using a sliding window that slides with the scan order, allowing context determination without needing to calculate relative locations and reducing data access, enabling efficient parallel context calculations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If position-based context support neighborhood is used, then coding accuracy is improved, but computational complexity increases

Engineering Contradiction:
Improvecoding accuracyVSAvoidcomputational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The context support neighborhood is segmented into a fixed-size sliding window that processes only a limited number of neighboring coefficients. This segmentation approach maintains coding accuracy by focusing on the most relevant neighbors while reducing computational complexity through bounded processing scope.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies local quality by using a sliding window that adapts to the local characteristics of transform coefficients. The context derivation is performed locally within the window boundaries, allowing accurate context modeling for each coefficient based on its immediate neighbors without requiring global position calculations.

Inventive Principle:
Principle #3Local quality

2Measurement precision

If position-based context support neighborhood is used, then coding accuracy is improved, but data access requirements increase

Engineering Contradiction:
Improvecoding accuracyVSAvoiddata access
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The sliding window segments the data access scope to a fixed, manageable size. Instead of accessing extensive data based on relative positions, the patent limits data access to coefficients within the sliding window boundaries, reducing memory access requirements while maintaining coding accuracy through focused local context analysis.

Inventive Principle:
Principle #1Segmentation

3Productivity

If parallel processing is implemented, then processing speed is improved, but context determination complexity increases

Engineering Contradiction:
Improveprocessing speedVSAvoidcontext determination complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The sliding window approach segments the context determination process into independent, fixed-size units that can be processed in parallel. Each coefficient within the sliding window can have its context determined independently using the same windowing logic, enabling parallel processing without increasing overall system complexity.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP2820845B1Scan-based sliding window in context derivation for transform coefficient coding
Publication Date: 2018.08.08 QUALCOMM INC
  • EP2820845B1 patent drawingFigure 1
  • EP2820845B1 patent drawingFigure 2
  • EP2820845B1 patent drawingFigure 3

AI summary

A video coding process that includes defining a context derivation neighborhood for one of a plurality of transform coefficients based on a transform coefficient scan order. The process also includes determining a context for the one of the plurality of transform coefficients based on the context derivation neighborhood. The process also includes coding the one of the plurality of transform coefficients based on the determined context.