Multi-Branch Image Enhancement Architecture with Regional Weight Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing deep learning networks for image enhancement struggle to achieve balanced and natural results when performing multiple enhancement processes simultaneously, due to architectural limitations and computational inefficiencies.
Innovation Solution
The proposed multi-branch processing architecture allows different types of image enhancements to be performed in parallel on separate branches, with regional weight control techniques to adjust enhancement effects based on image frequency characteristics, reducing computational burden and improving enhancement quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple deep learning networks are used for different enhancement processing, then different enhancement effects can be achieved, but computing time increases due to multiple inferences
Solution Approach 1:
The patent merges multiple separate deep learning networks into a single unified network that can perform multiple enhancement processing types (super-resolution, de-noising, de-blurring, sharpening) simultaneously through different processing branches, eliminating the need for multiple separate inferences and reducing computing time
2Adaptability or versatility
If multiple processing blocks are attached to a main network for different enhancement processing, then different enhancement effects can be achieved, but enhancements cannot be performed at the same time due to dependency
Solution Approach 1:
The patent segments the single deep learning network into multiple independent processing branches, where each branch is responsible for a specific enhancement processing type. This segmentation allows each branch to operate independently and simultaneously without dependency on other branches, enabling parallel processing of multiple enhancement operations
3Manufacturing precision
If image enhancement is performed on full-size images, then processing quality is maintained, but computational burden increases
Solution Approach 1:
The patent applies different processing resolutions to different processing branches based on their specific requirements. The super-resolution branch processes images at full size to maintain quality, while other enhancement branches (de-noising, de-blurring, sharpening) process down-sampled images at lower resolutions, reducing computational burden for operations that do not require full-size processing
Data Source
AI summary
An image processing system includes: a first image processing device for performing a first image enhancement process on a source image to generate a first enhanced image; one or more second images processing device, each of which is used to perform a second image enhancement processing on a size-reduced image generated based on the source image, and accordingly to output one or more second enhanced images whose size identical to the source image; and an output controller for analyzing regional frequency characteristics of the source image to generate an analysis result, determining one or more region weights according to the analysis result, and synthesize the first enhanced image with the one or more second enhanced images according to the one or more region weights, thereby to generate an output image.