Chroma Block Merging for Video Decoder Hardware Complexity Reduction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing video compression methods face challenges in efficiently coding chroma samples, particularly in reducing hardware complexity and achieving compact representation, due to the generation of chroma prediction blocks with sizes less than 4 pixels along each side, which is suboptimal for hardware efficiency.

Innovation Solution

The method involves merging chroma blocks with dimensions less than a threshold into larger blocks with at least 4 pixels along each side for prediction, using a decoder to derive prediction information and generate reconstructed chroma samples, thereby reducing hardware complexity and improving representation efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If chroma blocks are kept at small sizes (less than 4 pixels along each side), then the representation precision is improved, but the device complexity increases

Engineering Contradiction:
Improvechroma sample representation precisionVSAvoidhardware complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent merges multiple small chroma blocks (each less than 4 pixels along each side) into a single larger merged block that is at least 4 pixels along each side. This merging operation reduces the number of separate prediction blocks, thereby reducing hardware complexity while maintaining adequate chroma representation precision through the merging of adjacent small blocks.

Inventive Principle:
Principle #5Merging (Combining)

2Device complexity

If chroma blocks are merged into larger blocks, then the device complexity is reduced, but the measurement precision deteriorates

Engineering Contradiction:
Improvehardware complexityVSAvoidchroma sample representation precision
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent merges multiple small chroma blocks into a larger merged block to reduce hardware complexity. The merging is performed on adjacent small blocks that collectively maintain adequate chroma representation, thus reducing the number of separate processing units while preserving necessary precision.

Inventive Principle:
Principle #5Merging (Combining)

3Manufacturing precision

If chroma prediction blocks with sizes less than 4 pixels along each side are generated, then the manufacturing precision is improved, but the ease of manufacture deteriorates

Engineering Contradiction:
Improvechroma block dimension precisionVSAvoidhardware implementation ease
Core Design Contradiction:
Manufacturing precisionVSEase of manufacture

Solution Approach 1:

The patent merges multiple small chroma blocks (less than 4 pixels along each side) into a larger merged block (at least 4 pixels along each side). This merging simplifies hardware implementation by reducing the number of separate prediction blocks and associated processing logic, while the precise merging of adjacent small blocks maintains adequate chroma representation precision.

Inventive Principle:
Principle #5Merging (Combining)

Data Source

PatentEP3619915B1Coding chroma samples in video compression
Publication Date: 2024.06.26 HUAWEI TECH CO LTD
  • EP3619915B1 patent drawingFigure 1A~1E
  • EP3619915B1 patent drawingFigure 2
  • EP3619915B1 patent drawingFigure 3~4C

AI summary

A decoder and a method to process chroma blocks by the decoder are described. The method includes receiving a coding block from an encoder; partitioning the coding block into a plurality chroma blocks; merging at least two chroma blocks among the plurality chroma blocks into a merged block, where a width or height of a chroma block among the at least two chroma blocks is less than a threshold, and a width or height of the merged block is no less than the threshold; deriving prediction information for the merged block; generating prediction samples according to the derived prediction information; and obtaining reconstructed chroma samples according to the generated prediction samples. By merging the at least two chroma blocks, whose width or height is less than a threshold, into the merged block, hardware complexity can be reduced.