Image Encoding with Shared Quantization for Palette Escapes

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image encoding methods, such as HEVC and VVC, face challenges in managing quantization parameters to prevent unnecessary increases in code amount and maintain image quality, particularly in palette encoding for artificially-created images.

Innovation Solution

An image encoding device that adaptively corrects quantization parameters based on whether transforming processing is applied, using reference values for blocks to be encoded, and employs palette mode with escape values to optimize encoding efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If the quantization parameter is set smaller than 4 to increase gradations, then image quality after compression is improved, but the amount of code increases unnecessarily

Engineering Contradiction:
Improveimage qualityVSAvoidamount of code
Core Design Contradiction:
Manufacturing precisionVSQuantity of substance

Solution Approach 1:

The patent dynamically adjusts the quantization parameter based on the block type (transformed or non-transformed) and encoding mode (palette or non-palette). For non-transformed blocks with palette mode, the quantization parameter is clamped to a reference value (typically 4) to prevent unnecessary code increase. This conditional parameter adjustment resolves the contradiction by applying different quantization strategies to different block types.

Inventive Principle:
Principle #35Parameter changes

2Productivity

If palette encoding is used for artificially-created images, then encoding efficiency is improved, but quantization parameter management becomes complex

Engineering Contradiction:
Improveencoding efficiencyVSAvoidquantization parameter management
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent applies different quantization parameter management strategies to different local regions (blocks) based on their encoding mode. For blocks using palette mode, the quantization parameter is constrained to a reference value, while other blocks can use optimized quantization parameters. This localized approach simplifies overall management by providing clear rules for specific block types while maintaining flexibility elsewhere.

Inventive Principle:
Principle #3Local quality

3Quantity of substance

If the quantization parameter is increased above 4, then the quantization step increases and compressed value decreases, but image quality deteriorates

Engineering Contradiction:
Improvecode amountVSAvoidimage quality
Core Design Contradiction:
Quantity of substanceVSManufacturing precision

Solution Approach 1:

The patent implements a feedback mechanism where the quantization parameter is adjusted based on the determined block characteristics (transformed/non-transformed, palette/non-palette mode). The system monitors the block type and automatically selects an appropriate quantization parameter range, preventing both excessive compression (QP>4) and unnecessary code increase (QP<4), thereby maintaining optimal image quality for each block type.

Inventive Principle:
Principle #23Feedback

Data Source

PatentUS12464119B2Image encoding device, image encoding method, and program, image decoding device, image decoding method, and non-transitory computer-readable storage medium
Publication Date: 2025.11.04 CANON KK
  • US12464119B2 patent drawing
  • US12464119B2 patent drawing
  • US12464119B2 patent drawing

AI summary

An encoding device, if it is determined that a transforming processing is applied to a block to be encoded, encodes the block using a first quantization parameter, if it is determined that the transforming processing is not applied to the block and the first quantization parameter is less than a reference value, encodes the block using the reference value as a quantization parameter, and, if it is determined that a pallet mode is applied to the block and the first quantization parameter is less than a reference value, encodes the block using the reference value as a quantization parameter for an escape value, wherein the reference value used if it is determined that the transforming processing is not applied to the block and the reference value used for the escape value are the same.