Adaptive Quantization Parameter Adjustment for Image Compression

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current image processing techniques face challenges in efficiently encoding and decoding images, particularly in adapting quantization parameters based on image content, which affects compression efficiency and quality.

Innovation Solution

A method and system that access and adjust quantization parameters using implicit rate control, where a quantization parameter is determined for encoding a sample, and if a flatness criterion is met, it is adjusted and explicitly indicated to the decoding system, allowing for adaptive encoding and decoding.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If adaptive quantization is applied based on image content, then compression efficiency is improved, but device complexity increases

Engineering Contradiction:
Improvecompression efficiencyVSAvoiddevice complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent applies different quantization parameters to different regions of the image based on local flatness characteristics. Smooth regions use coarser quantization while detailed regions use finer quantization, optimizing compression efficiency for each region's specific content requirements.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The system uses decoder-side computation to derive motion vectors and flatness indicators without requiring additional encoder complexity. The decoder performs template matching and flatness analysis using already-decoded pixel data, making the adaptive quantization self-sufficient at the decoding stage.

Inventive Principle:
Principle #25Self-service

2Measurement precision

If explicit quantization parameter transmission is used, then decoding accuracy is improved, but bandwidth usage increases

Engineering Contradiction:
Improvedecoding accuracyVSAvoidbandwidth usage
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent extracts only the essential flatness indication information from the image content and transmits this compact representation to the decoder. Instead of transmitting full quantization parameter maps, only the derived flatness indicators are sent, significantly reducing bandwidth requirements while maintaining decoding accuracy.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The system transforms the image content into a different parameter space by computing flatness indicators from pixel values. This parameter transformation allows the decoder to derive appropriate quantization parameters from the compact flatness information without transmitting the full parameter set.

Inventive Principle:
Principle #35Parameter changes

3Productivity

If decoder-side motion vector derivation is used, then coding efficiency is improved, but computational load increases

Engineering Contradiction:
Improvecoding efficiencyVSAvoidcomputational load
Core Design Contradiction:
ProductivityVSPower

Solution Approach 1:

The patent performs template matching and flatness analysis only on a subset of pixels rather than the entire image. By processing only the necessary regions for motion vector derivation and quantization parameter selection, the computational load is reduced while maintaining coding efficiency improvements.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP2940998B1Bounded rate near-lossless and lossless image compression
Publication Date: 2020.09.02 AVAGO TECHNOLOGIES INTERNATIONAL SALES PTE LTD
  • EP2940998B1 patent drawingFigure 1
  • EP2940998B1 patent drawingFigure 2
  • EP2940998B1 patent drawingFigure 3

AI summary

A system adjusts the quantization parameter used to encode an image sample (e.g., pixel component). The system accesses a quantization parameter for encoding a current image pixel from an implicit rate control and obtains a measure of another image sample within a particular distance from the sample. When the measure meets a predetermined criterion, the system adjusts the quantization parameter, encodes the pixel after adjusting the quantization parameter, and explicitly indicates the adjusting of the quantization parameter to a decoding system.