Image Block Division Mode Determination for Video Coding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing video coding and decoding standards fail to accurately consider image texture information and object boundaries, leading to inaccurate motion estimation and prediction results due to the use of rectangular blocks that do not align with irregular image content.

Innovation Solution

A method that determines a division mode for image blocks by calculating a second length based on the direction of a division line, a scale value, and a position parameter value, allowing for non-square division of image blocks into irregular shapes such as triangles or pentagons, eliminating the need for fitting a division line and simplifying the identification process.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If rectangular blocks are used for motion estimation, then the coding process is simple, but the prediction accuracy is insufficient due to not aligning with irregular image content

Engineering Contradiction:
Improvecoding simplicityVSAvoidprediction accuracy
Core Design Contradiction:
Ease of manufactureVSMeasurement precision

Solution Approach 1:

The patent divides an image block into multiple triangular sub-blocks using division lines with different directions (first direction and second direction). This segmentation allows the image to be represented by irregular shapes that better align with object boundaries and texture information, improving prediction accuracy while maintaining manageable coding complexity through systematic division rules.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces asymmetric triangular divisions by using two different division line directions, creating non-uniform sub-blocks that adapt to the local image content. This asymmetry enables better fitting of irregular image structures compared to symmetric rectangular blocks, thereby improving motion estimation accuracy.

Inventive Principle:
Principle #4Asymmetry

2Measurement precision

If non-square division method is used, then the prediction accuracy is improved, but the device complexity increases due to additional identification requirements

Engineering Contradiction:
Improveprediction accuracyVSAvoididentification complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent applies different division line directions (first direction or second direction) to different regions or contexts within the image block. By selecting appropriate division directions based on local image characteristics, the method achieves high prediction accuracy while managing complexity through localized rather than global complex identification.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

Instead of identifying complex geometric parameters (angles, distances) to define division lines, the patent inverts the approach by using simple directional indicators (first direction or second direction) to define triangular divisions. This inversion simplifies the identification process while maintaining the ability to represent irregular image content effectively.

Inventive Principle:
Principle #13The other way round (Inversion)

3Ease of operation

If rectangular blocks are used, then the coding process is straightforward, but texture information and object boundaries are not properly considered

Engineering Contradiction:
Improvecoding process simplicityVSAvoidtexture and boundary information
Core Design Contradiction:
Ease of operationVSLoss of information

Solution Approach 1:

By segmenting the image block into triangular sub-blocks using division lines, the patent creates smaller, more granular regions that can better capture local texture variations and object boundaries. This segmentation preserves fine-grained image information that would be lost in larger rectangular blocks, while the systematic division approach keeps the coding process manageable.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP2658256B1Method and device for image encoding and decoding
Publication Date: 2017.04.26 HUAWEI TECH CO LTD
  • EP2658256B1 patent drawingFigure 1
  • EP2658256B1 patent drawingFigure 2A
  • EP2658256B1 patent drawingFigure 2B

AI summary

Embodiments of the present invention provide a method and a device for coding and decoding images. The method for coding and decoding images includes: determining a second length according to a direction of a division line, a scale value, and a position parameter value, where the second length is a distance from a point existent on a boundary of an image block and corresponding to a point of the image block to an endpoint of the division line, where the endpoint of the division line is an intersection point of the division line and the boundary of the image block or an extension line of the boundary, and the position parameter value represents a relative position of the endpoint of the division line against a reference endpoint in a reference direction; determining a division mode of the image block according to the direction of the division line, a first length, the second length, and the scale value; coding the image block according to the division mode of the image block, where the first length is a distance from the point of the image block to the boundary of the image block; and coding a direction parameter value, the scale value, and the position parameter value. With the technical solution provided in the embodiments of the present invention, the determining of the division mode of the image block on the image coder is simplified.