Adaptive Quantization Parameter Adjustment Using JND Thresholds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image and video compression methods face challenges in optimizing quantization parameters to balance objective and subjective quality, often leading to reduced compression efficiency due to limitations in adapting to human visual characteristics.
Innovation Solution
A method and apparatus for adjusting quantization parameters using initial values based on image objective quality evaluation indices and just noticeable difference (JND) thresholds, which determine block JND thresholds and adjust quantization parameters to reduce subjective redundancy while maintaining objective quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional quantization methods are used based on objective quality evaluation indices, then the compression rate can be improved, but the subjective quality perception deteriorates due to insufficient adaptation to human visual characteristics
Solution Approach 1:
The patent divides the image into multiple pixel blocks and calculates different JND thresholds for each block based on local luminance and contrast characteristics. This allows the quantization parameter to be adaptively adjusted for each local region, preserving visually sensitive areas while allowing more aggressive compression in less sensitive areas, thereby improving compression rate without significantly degrading subjective quality
Solution Approach 2:
The patent introduces JND threshold as a new parameter to adjust the quantization parameter dynamically. By calculating JND thresholds based on local image characteristics (luminance, contrast, spatial frequency) and using them to modulate the quantization parameter, the system adapts to human visual characteristics, achieving better compression efficiency while maintaining perceived quality
2Manufacturing precision
If adaptive quantization based on JND thresholds is implemented, then subjective quality is preserved, but the computational complexity increases due to additional JND threshold calculations
Solution Approach 1:
The patent segments the image into multiple pixel blocks and processes each block independently to calculate JND thresholds. This segmentation approach allows parallel computation and reduces the overall computational burden compared to processing the entire image as a single unit, while still capturing local visual characteristics for accurate quality preservation
Solution Approach 2:
The patent calculates JND thresholds selectively based on key local characteristics (luminance, contrast, spatial frequency) rather than performing exhaustive analysis of all image features. This partial action approach achieves sufficient adaptation to human visual characteristics without the prohibitive computational cost of complete visual model simulation
Data Source
AI summary
An implementation of a method for adjusting a quantization parameter for adaptive quantization may include: acquiring at least one pixel block corresponding to a to-be-compressed image and an initial quantization parameter adjustment value corresponding to the pixel block, the initial quantization parameter adjustment value being generated based on an image objective quality evaluation index; determining a just noticeable difference, JND, threshold corresponding to each pixel point in the at least one pixel block; determining a block JND threshold corresponding to each pixel block, based on the JND threshold corresponding to each pixel point in the at least one pixel block; and adjusting the corresponding initial quantization parameter adjustment value based on the block JND threshold to generate an adjusted quantization parameter adjustment value corresponding to the at least one pixel block.