Adaptive Image Fusion Using Local Attribute Differences
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image processing methods for post-enhancement produce poor enhancement effects due to global fusion techniques that fail to account for localized differences in image attributes.
Innovation Solution
An image processing method that involves obtaining attribute characterizations of a source and enhanced image, comparing these to determine local and enhanced fusion weights, and fusing the images based on these weights to generate a fused image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If global fusion is performed on source image and enhanced image based on a designed fusion factor, then the fusion process is simple and fast, but the enhancement effect is poor and lacks adaptability
Solution Approach 1:
The patent segments the image into different regions (edge regions and non-edge regions) and applies different fusion strategies to each region. Edge regions use gradient information to determine fusion weights, while non-edge regions use luminance difference information, thereby achieving regional adaptability in the fusion process
Solution Approach 2:
The patent implements local quality by calculating different fusion weights for different pixel positions based on local image characteristics. Each pixel's fusion weight is determined by local gradient magnitude and luminance difference, making the fusion process adaptive to local image features rather than applying a global fixed factor
2Device complexity
If fixed fusion factors are used for all pixels, then the fusion process is simple, but the enhancement effect is poor across different image regions
Solution Approach 1:
The patent transforms the static fixed fusion factor into a dynamic adaptive weight that changes according to local image characteristics. The fusion weight for each pixel is dynamically calculated based on gradient magnitude and luminance difference, allowing the system to adapt to varying enhancement needs across different regions
Solution Approach 2:
The patent changes the fusion parameter from a fixed global factor to variable local weights that are calculated based on image attributes. By using gradient magnitude and luminance difference as basis for weight calculation, the system dynamically adjusts fusion parameters to match local image characteristics
3Productivity
If conventional enhancement methods are applied uniformly, then the processing is efficient, but details are lost in certain regions
Solution Approach 1:
The patent segments the image processing into edge region handling and non-edge region handling. Edge regions are processed using gradient-based fusion to preserve structural details, while non-edge regions use luminance-based fusion to preserve texture and detail information, thereby preventing detail loss in both regions
Solution Approach 2:
The patent incorporates feedback mechanisms by using gradient magnitude and luminance difference information from the enhanced image to dynamically adjust fusion weights. This feedback loop ensures that regions where enhancement may cause detail loss receive appropriate weight adjustment to preserve important image features
Data Source
AI summary
An image processing method and apparatus, the method includes obtaining an attribute characterization of a source image and an attribute characterization of an enhanced image, the enhanced image being obtained by enhancing the source image, comparing the attribute characterization of the source image with the attribute characterization of the enhanced image to obtain attribute differences between at least part of pixel positions in the source image and corresponding at least part of pixel positions in the enhanced image, determining local fusion weights of the at least part of pixel positions in the source image based on the attribute differences and the attribute characterization of the enhanced image, determining enhanced fusion weights of the at least part of pixel positions in the enhanced image, and fusing the source image and the enhanced image to obtain a fused image.