Image Processing Apparatus Dynamic Parameter Adjustment for Memory Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing technologies face challenges in improving image quality in response to variable image quality due to factors like transmission medium, resolution, and bit rate, leading to increased memory usage when using different machine learning networks for each image quality level.
Innovation Solution
An image processing apparatus that adjusts parameters based on the compression rate of each frame, using pre-stored parameters for known rates and calculating new parameters for unknown rates through linearity analysis, to optimize image quality improvement while minimizing memory consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If different machine learning networks are applied according to the degree of quality of an image in response to a variable image quality, then image quality improvement performance is enhanced, but memory consumption increases in proportion to the range of image quality
Solution Approach 1:
A single machine learning network is designed to handle multiple image quality levels through dynamic parameter adjustment rather than requiring separate specialized networks for each quality level. The network universally processes images across variable compression rates by adapting its parameters in real-time.
Solution Approach 2:
The patent dynamically changes parameters of the machine learning network based on the compression rate of input frames. By adjusting parameters according to the detected image quality level, the system optimizes performance for each quality range without needing separate networks, thereby reducing memory consumption while maintaining improvement performance.
2Measurement precision
If parameters are pre-stored for multiple compression rates, then image processing accuracy is improved, but memory usage increases
Solution Approach 1:
The system pre-stores parameters only for representative compression rates (e.g., first, second, and third compression rates) rather than all possible rates. This preliminary preparation of key parameters enables accurate processing for common cases while minimizing storage requirements.
Solution Approach 2:
For compression rates that do not have pre-stored parameters, the system uses linear interpolation between adjacent known compression rates to calculate appropriate parameters. This intermediary calculation method maintains processing accuracy for unknown rates without requiring additional parameter storage.
3Quantity of substance
If a single machine learning network is used for all image quality levels, then memory consumption is reduced, but image quality improvement performance decreases
Solution Approach 1:
The patent transforms a static single-network approach into a dynamic system where the network's parameters are continuously adjusted based on the input image's compression rate. This dynamic adaptation allows one network to effectively handle variable image quality levels, maintaining performance while reducing memory usage compared to multiple static networks.
Data Source
AI summary
An image processing apparatus is provided. The image processing apparatus includes a processor configured to, in response to an image including a plurality of frames being input, change a predetermined parameter to a parameter corresponding to a compression rate of each of the plurality of frames for each frame, and process the input image by using the parameter changed for each frame, and an output interface configured to output the processed image.