Image Decoding With Gradient-Adaptive Pixel QP Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding and decoding technologies suffer from significant quantization distortion due to applying the same quantization parameter (QP) to all pixel points within a coding unit, leading to reduced decoding accuracy and authenticity.
Innovation Solution
Determine a predicted residual value for each pixel point based on its gradient, adjusting the quantization parameter (QP) accordingly to perform dequantization on a pixel-by-pixel basis, reflecting the complexity of each pixel point.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If the same quantization parameter (QP) is applied to all pixel points in a coding unit, then the coding and decoding process is simple, but quantization distortion increases and decoding accuracy decreases
Solution Approach 1:
The patent applies different quantization parameters to different pixel points within the same coding unit based on their local characteristics (gradient values). This allows regions with high frequency content to use smaller QP values for better preservation of details, while regions with low frequency content use larger QP values for better compression, thereby reducing overall quantization distortion without uniformly increasing complexity across the entire block.
Solution Approach 2:
The patent segments the coding unit into multiple regions or pixel points and applies adaptive quantization parameters to each segment based on local gradient characteristics. This segmentation approach enables differentiated treatment of different regions, improving decoding accuracy by preserving important local features while maintaining reasonable compression efficiency.
2Productivity
If the same quantization parameter (QP) is applied to all pixel points in a coding unit, then the compression process is efficient, but quantization distortion increases and picture quality deteriorates
Solution Approach 1:
The patent calculates gradient values for each pixel point and uses these local characteristics to determine adaptive quantization parameters. This allows the compression process to be more efficient in smooth regions (using larger QP) while preserving important edges and details in complex regions (using smaller QP), thereby reducing quantization distortion without significantly compromising compression efficiency.
Solution Approach 2:
The patent introduces dynamic quantization parameter adjustment based on local image characteristics (gradient values). Instead of using a static QP for the entire coding unit, the QP varies dynamically across different pixel points according to their local complexity, achieving better balance between compression efficiency and picture quality.
Data Source
AI summary
A picture decoding method is provided, including: determining a predicted residual value of a pixel point according to a prediction manner of the pixel point, where the predicted residual value is configured to reflect a gradient of the pixel point, the prediction manner is configured to indicate a position of one or more reconstructed pixel points referenced when performing prediction on a pixel point, and the pixel point is any pixel point in a current coding block; determining a target quantization parameter (QP) value of a first pixel point according to the predicted residual value of the first pixel point, where the first pixel point is a target pixel point in the current coding block, and the target pixel point is a preset pixel point for adjusting the QP value; and performing dequantization on the first pixel point according to the target QP value of the first pixel point.