Video Encoder Block Size Selection via Decision Tree

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Evaluating different block sizes for video encoding in HEVC is costly, particularly when fast encoding is required, as it significantly increases encoding time due to the need to assess various transform and prediction unit sizes.

Innovation Solution

A method and encoder that quickly choose block sizes by comparing predefined combinations of block sizes, focusing on the most efficient options first, such as comparing the first combination against the second and then against the third, to select the best option based on distortion measures, thereby reducing the need for full evaluations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If different block sizes are evaluated to improve compression efficiency, then compression efficiency is improved, but encoding time increases significantly

Engineering Contradiction:
Improvecompression efficiencyVSAvoidencoding time
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The patent segments the evaluation process by dividing block size selection into multiple stages: initial selection of a limited set of candidate block sizes, preliminary evaluation of these candidates, and final selection based on distortion measures. This segmentation avoids evaluating all possible block sizes while still achieving good compression efficiency.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies partial action by evaluating only a subset of block size options rather than all possible sizes. It performs sufficient evaluation to achieve good compression results without the excessive action of exhaustively testing every block size combination, thereby reducing encoding time while maintaining effectiveness.

Inventive Principle:
Principle #16Partial or excessive action

2Manufacturing precision

If full evaluation of all block size combinations is performed to achieve optimal compression, then compression quality is improved, but device complexity increases

Engineering Contradiction:
Improvecompression qualityVSAvoidencoder complexity
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The encoder complexity is reduced by segmenting the block size selection into manageable stages with predetermined candidate sets. This avoids the need for a complex exhaustive evaluation system while still achieving optimal compression quality through structured, multi-stage decision-making.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent changes the parameter of block size evaluation by using predetermined candidate sets and distortion measures instead of exhaustive evaluation. This parameter change simplifies the encoder's decision-making process while maintaining the ability to achieve high compression quality.

Inventive Principle:
Principle #35Parameter changes

3Productivity

If multiple transform unit sizes are evaluated to improve compression efficiency, then compression efficiency is improved, but encoding time increases

Engineering Contradiction:
Improvecompression efficiencyVSAvoidtransform unit evaluation time
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The patent applies partial action by evaluating only necessary transform unit size options based on the selected block size and content characteristics. This avoids excessive evaluation of all possible transform sizes while achieving sufficient compression efficiency for practical applications.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP3053337B1Encoder and method for video processing
Publication Date: 2020.07.29 TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)
  • EP3053337B1 patent drawingFigure 1
  • EP3053337B1 patent drawingFigure 2A~2D
  • EP3053337B1 patent drawingFigure 3

AI summary

The embodiments relate to an encoder for encoding a bitstream representing a sequence of pictures of a video stream. The encoder is operative to divide the pictures into blocks of different types, such as coding units CUs, prediction units PUs, and transform units TUs, which are associated with each other. Two sizes are possible for the blocks and the possible combinations are restricted to three cases: 1) all blocks are of large size, 2) the CUs and PUs have large size, while associated TUs have small size, 3) all blocks are of small size. The possible combinations are defined for a part of a picture and compared against each other using a decision tree. In this way an advantageous block size configuration for encoding the bitstream is determined at a reduced computational cost.