Picture Decoding with Quantization Groups to Reduce Distortion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Large quantization distortion in picture encoding and decoding processes in video compression technologies.
Innovation Solution
A decoding method that involves analyzing a code stream to determine quantization parameter (QP) values for coding units (CUs) by dividing them into quantization groups (QGs) based on specific rules, such as coordinate or proportion thresholds, and adaptively adjusting QP values for pixels based on surrounding reconstructed pixels, while encoding methods involve similar QP determination for efficient decoding and encoding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If uniform quantization parameter is used for entire coding unit, then encoding complexity is reduced, but quantization distortion increases
Solution Approach 1:
The coding unit is divided into multiple quantization groups (QGs), where each QG can have its own independently determined quantization parameter. This segmentation allows different regions to use optimized QP values based on their specific characteristics, reducing overall quantization distortion while maintaining manageable encoding complexity through structured division.
Solution Approach 2:
Different quantization parameters are assigned to different quantization groups within the coding unit based on local characteristics such as coordinate thresholds, proportion thresholds, or visual importance. This local optimization ensures that each region receives the appropriate QP value for its specific requirements, balancing compression efficiency with quality preservation.
2Manufacturing precision
If multiple quantization parameters are assigned to different quantization groups, then quantization distortion is reduced, but determining complexity increases
Solution Approach 1:
The patent establishes predetermined rules for determining quantization parameters in advance, such as coordinate threshold rules or proportion threshold rules. These pre-defined guidelines simplify the QP determination process by providing clear criteria that can be automatically applied during encoding, reducing the computational burden despite using multiple QP values.
Solution Approach 2:
The patent introduces flexible parameter adjustment mechanisms where quantization parameters can be modified based on various conditions including coordinate sums, proportions, or visual characteristics. This allows the system to adapt QP values dynamically while maintaining controlled complexity through structured parameter change rules.
3Measurement precision
If quantization parameter is optimized for each pixel, then decoding performance is improved, but processing time increases
Solution Approach 1:
The pixel-level optimization is achieved through segmentation into quantization groups, where each group can have its own QP value. This approach provides pixel-level precision control while avoiding the computational burden of completely independent pixel-by-pixel processing, as groups share common characteristics and can be processed together.
Solution Approach 2:
Instead of fully optimizing every single pixel independently (excessive action), the patent applies optimization at the quantization group level, which provides sufficient precision for most applications while significantly reducing processing time. This partial optimization approach achieves diminishing returns beyond a certain group-level granularity.
Data Source
AI summary
A decoding method and apparatus, and an encoding method and apparatus for a picture are provided, and relate to the field of video encoding and decoding. The decoding method includes: analyzing a code stream to obtain one or more picture frames, wherein a picture frame comprises one or more coding units (CUs); determining a plurality of quantization parameter (QP) values for the picture frame, wherein a CU comprises a plurality of quantization groups (QGs), and a QG corresponds to a QP value; and decoding the picture frame based on the plurality of QP values.