Video Processing Apparatus Block Size Selection at Picture Boundaries

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional video encoding techniques require neighbor pixel padding at picture boundaries for intra prediction, which can lead to inefficiencies and inaccuracies, especially when transform blocks cross picture boundaries.

Innovation Solution

A video processing apparatus and method that determines picture boundary information to select block sizes, ensuring transform units and mode information units are not across picture boundaries, thereby avoiding the need for padding by constraining sizes to be integer multiples of superblock side lengths.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If transform units are allowed to cross picture boundaries for flexible block-based coding, then coding flexibility is improved, but neighbor pixel padding is required at picture boundaries which increases complexity and reduces accuracy

Engineering Contradiction:

Improvecoding flexibilityVSAvoidpadding operation complexity

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The picture is segmented into superblocks with boundaries aligned to picture boundaries. Transform units are constrained to be within superblocks, which are further constrained to not cross picture boundaries. This hierarchical segmentation ensures that no transform unit needs to access pixels outside the picture boundary, eliminating the need for padding operations while maintaining flexible block-based coding within the constrained structure.

Inventive Principle:

Principle #1Segmentation

2Adaptability or versatility

If transform units are allowed to cross picture boundaries, then block-based coding flexibility is improved, but intra prediction accuracy deteriorates due to required neighbor pixel padding

Engineering Contradiction:

Improveblock-based coding flexibilityVSAvoidintra prediction accuracy

Core Design Contradiction:

Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The picture is divided into superblocks that are constrained to align with picture boundaries. Transform units are further segmented within superblocks. This segmentation hierarchy ensures that transform units performing intra prediction never require neighbor pixels outside the picture boundary, eliminating padding-related accuracy loss while preserving coding flexibility within the segmented structure.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent applies different constraints to different regions: superblocks and transform units near picture boundaries are constrained to not cross boundaries, while internal blocks maintain full flexibility. This local quality approach ensures high prediction accuracy at boundaries without unnecessarily restricting coding flexibility in internal regions.

Inventive Principle:

Principle #3Local quality

3Productivity

If block sizes are constrained to not cross picture boundaries, then padding operations are eliminated improving efficiency, but block size flexibility is reduced

Engineering Contradiction:

Improveencoding efficiencyVSAvoidblock size flexibility

Core Design Contradiction:

ProductivityVSAdaptability or versatility

Solution Approach 1:

The patent introduces superblocks as an intermediate segmentation layer between picture boundaries and transform units. This allows transform units to have flexible sizes and positions within superblocks, while superblocks themselves are constrained to align with picture boundaries. This hierarchical segmentation eliminates padding operations (improving efficiency) while preserving transform unit size flexibility within the superblock structure.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent dynamically adjusts block sizes and positions within superblocks based on content characteristics, while maintaining the constraint that superblocks align with picture boundaries. This dynamic adaptation allows the encoder to optimize for efficiency (avoiding padding) while retaining sufficient flexibility to adapt to different picture contents through variable block sizing within the constrained superblock framework.

Inventive Principle:

Principle #15Dynamics

Data Source

PatentUS10070070B2Video processing apparatus with transform unit size selection, mode information unit size selection and/or picture width/height decision, and related video processing method thereof

Publication Date: 2018.09.04 XUESHAN TECH INC

AI summary

One exemplary video processing apparatus includes a control circuit and a size selection circuit. The control circuit determines picture boundary information. The size selection circuit refers to at least the picture boundary information to select a size for a block associated with encoding of a picture, wherein selection of the size is constrained by the picture boundary information to ensure that the block with the selected size is not across a picture boundary of the picture.