Non-Uniform Super-Resolution Image Processing via Quality Map Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing technologies are not optimized for machine learning tasks, which require improved image encoding and decoding performance, and struggle with handling images of non-uniform quality.
Innovation Solution
A non-uniform super-resolution method that separates input images into high-quality and low-quality regions based on a quality map, converts the high-quality regions to match the low-quality regions, and combines them for processing through a super-resolution network, using techniques like masking, spatial attention, and quality map reflection to enhance image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing image processing technology is used, then human viewing quality is maintained or improved, but machine learning task performance is not optimized
Solution Approach 1:
The patent applies local quality by differentiating treatment between high-quality and low-quality regions of images. The system separates images into these regions based on quality maps and applies super-resolution processing selectively to low-quality regions while preserving high-quality regions, thereby optimizing image data for machine learning tasks without uniformly degrading all image areas.
Solution Approach 2:
The patent segments the image processing task into distinct modules: quality map generation, image separation into high/low quality regions, selective super-resolution processing, and recombination. This segmentation allows each component to be optimized for its specific function, improving overall performance for machine learning applications.
2Manufacturing precision
If uniform super-resolution is applied to entire images, then processing simplicity is maintained, but image quality degradation in certain regions cannot be addressed
Solution Approach 1:
The system applies different processing quality levels to different regions. High-quality regions are preserved as-is, while low-quality regions undergo super-resolution processing. This local differentiation improves overall image quality while avoiding unnecessary processing complexity in already high-quality areas.
Solution Approach 2:
Instead of applying full super-resolution processing to entire images (excessive action), the system applies processing only to the necessary low-quality regions (partial action). This reduces computational complexity while still achieving the desired quality improvement where needed.
3Manufacturing precision
If quality map is used to guide super-resolution, then image quality improvement is enhanced, but additional processing steps are required
Solution Approach 1:
The system performs preliminary actions by generating quality maps and separating images into high/low quality regions before applying super-resolution processing. These preliminary steps enable targeted processing that improves image quality while managing computational complexity through efficient workflow organization.
Data Source
AI summary
A non-uniform super-resolution method, device, and recording medium of an image of the present disclosure may include an image separation unit which separates an input image into a high-quality region and a low-quality region based on a quality map image, an image quality conversion unit which acquires a converted high-quality region by converting image quality of the high-quality region, and a first image combination unit which combines the converted high-quality region and the low-quality region to transmit them to a super-resolution network.