Template-Based Intra Prediction for Video Coding Efficiency

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The complexity and coding efficiency associated with template-based Intra prediction in video coding, particularly in High Efficiency Video Coding (HEVC), need to be reduced to improve performance.

Innovation Solution

The method involves determining whether to apply template-based Intra prediction to a current block based on its size, using a template matching candidate set to derive a target Intra mode or Intra-mode candidate set, and encoding or decoding the block with a selected mode from this set, while adjusting the number of Intra modes searched based on block size and structure.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If template-based Intra prediction is applied to all blocks, then coding efficiency is improved, but computational complexity increases

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

Solution Approach 1:

The patent applies template-based Intra prediction selectively based on block size characteristics. Specifically, it differentiates between small blocks (4×4, 8×4, 4×8) and larger blocks (8×8 and above), applying the complex template matching process only to blocks where it provides significant benefit. This local differentiation maintains coding efficiency for suitable blocks while reducing overall computational complexity by avoiding unnecessary processing on small blocks.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent modifies the Intra prediction process by changing the parameter of block size threshold. It establishes that template-based Intra prediction is applied only when the block size is 8×8 or larger, fundamentally altering the operational parameters of the prediction process to balance complexity and efficiency.

Inventive Principle:
Principle #35Parameter changes

2Measurement precision

If the number of Intra modes searched is increased, then prediction accuracy is improved, but processing time increases

Engineering Contradiction:
Improveprediction accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent applies different numbers of Intra prediction modes based on block size. For blocks of 8×8 or larger, it searches all 35 Intra prediction modes to ensure high accuracy. For smaller blocks (4×4, 8×4, 4×8), it limits the search to only 9 modes, significantly reducing processing time while maintaining adequate prediction quality for smaller regions.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent implements partial action by searching a complete set of 35 modes only when necessary (for larger blocks), and using a reduced set of 9 modes for smaller blocks where full mode search would be excessive. This selective approach optimizes the balance between exhaustive search for accuracy and limited search for efficiency.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentUS10230961B2Method and apparatus for template-based intra prediction in image and video coding
Publication Date: 2019.03.12 XUESHAN TECH INC
  • US10230961B2 patent drawing
  • US10230961B2 patent drawing
  • US10230961B2 patent drawing

AI summary

A method and apparatus for video coding using template-based Intra prediction are disclosed. According to one method, where determining whether to apply the template-based Intra prediction or one or more parameters associated with the template-based Intra prediction depends on the current block size. According to yet another method, the quad-tree plus binary tree (QTBT) structure is used to partition an image or one or more areas of the current image into blocks. If the template-based Intra prediction is used for a current block and the current block is non-square, the width and height of the L-shaped reference pixel line are determined according to width and height of the current block. The L-shaped reference pixel line comprises a top reference pixel segment above the top template and a left reference pixel segment adjacent to a left side of the left template.