Image Processing Using Composite Images for Data-Efficient Quality Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The management of large volumes of image data, especially in edge devices, is inefficient, leading to increased storage needs, resource wastage, and noise in machine learning models due to redundant and low-quality image additions.
Innovation Solution
A method for determining composite images from a group of similar images and assessing their quality to decide on retention or discard, using data distillation and capsule network models to optimize image storage and utilization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If all acquired image data is retained for model training, then the model has access to more training data, but storage space is excessively consumed and low-quality images introduce noise
Solution Approach 1:
The patent applies parameter changes by transforming the image data from its original form into composite images through data distillation. Multiple original images are processed and transformed into a single composite image that preserves essential training information while reducing storage requirements. This parameter transformation resolves the contradiction by changing the state of the data from individual images to compressed composite representations.
Solution Approach 2:
The patent uses copying by creating composite images that serve as simplified representations or copies of multiple original images. Instead of storing all original images, the system creates distilled composite copies that contain the essential training information. This allows the model to learn from the same information with reduced storage consumption.
2Loss of energy
If composite images are used to reduce storage, then storage efficiency improves, but image quality assessment becomes more complex
Solution Approach 1:
The patent applies preliminary action by performing data distillation and composite image generation before the quality assessment step. By pre-processing the images into composite form and pre-calculating quality metrics during the distillation process, the system simplifies the subsequent quality assessment. The quality evaluation leverages information already extracted during composite image creation, reducing the complexity of the assessment process.
3Loss of energy
If data distillation is applied to compress images, then storage needs are reduced, but processing time increases
Solution Approach 1:
The patent applies merging by combining multiple processing operations into a unified data distillation framework. Instead of separately compressing images and then assessing quality, the system merges these operations so that composite images are generated and quality metrics are calculated in an integrated process. This merging reduces overall processing time by eliminating redundant computation steps.
Data Source
AI summary
Embodiments of the present disclosure relate to a method, a device, and a computer program product for image processing. The method includes determining a group of composite images, a composite image in the group of composite images being composited from images of the same type in a plurality of images. The method further includes determining an image quality of an acquired target image based on the group of composite images. The method further includes processing the acquired target image based on the image quality. Through the method, the storage and management of image data are improved, the effective utilization rate of data is improved, and the consumption of resources is reduced.