Multi-Frame Image Fusion Using Local Weights to Suppress Halo
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The fusion of multi-frame exposure images faces challenges such as moving scenes, halo generation, and contrast compression, especially in complex light sources or backlit scenes, leading to a difference between human vision and captured images due to the limitations of sensor modules in capturing a full dynamic range.
Innovation Solution
A multi-frame image fusion method that introduces local information using light source diffusion weight, performing image alignment based on exposure sharpness, calculating fusion regions with local light source diffusion weights, and compressing the fused image to a preset dynamic range, while using algorithms like Laplacian pyramid fusion to enhance image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Illumination intensity
If multi-frame exposure images are fused to capture full dynamic range, then the dynamic range of the image is improved, but halos are generated around light sources
Solution Approach 1:
The patent applies local quality by introducing light source diffusion weight information in the spatial domain to specifically address the halo problem around light sources while preserving the benefits of multi-frame fusion. This localized approach allows different parts of the image (light source regions vs. non-light source regions) to be processed with different weights, eliminating halos in critical areas while maintaining dynamic range enhancement elsewhere.
Solution Approach 2:
The patent changes parameters by incorporating local light source diffusion weight information into the fusion process. By adjusting the weight parameters based on spatial domain analysis of light source characteristics, the system dynamically controls the contribution of each frame in different regions, thereby suppressing halos while preserving dynamic range benefits.
2Loss of information
If traditional multi-frame fusion algorithms are used, then image fusion is achieved, but contrast compression occurs and aesthetic quality deteriorates
Solution Approach 1:
The patent improves image quality by applying local quality principles through spatial domain analysis. By identifying light source regions and assigning appropriate diffusion weights, the system preserves local contrast and aesthetic appearance in critical areas while maintaining overall fusion completeness. This localized processing prevents contrast compression that occurs in traditional global fusion approaches.
Data Source
AI summary
A multi-frame image fusion method and system, an electronic device, and a storage medium are provided including performing image alignment based on a frame with highest exposure sharpness among a plurality of frames of images, wherein the image alignment includes extracting feature points of each of the plurality of frames to make the plurality of frames have a common spatial layout; calculating a fusion region of each frame based on local light source diffusion weight information in a spatial domain; performing image fusion on the plurality of frames with different exposures based on the fusion region, wherein the image fusion includes fusing the plurality of frames with different exposures into a fused image; and compressing the fused image to a preset dynamic range section. During the image fusion process, the local information is introduced into images to make the fused image possess a higher dynamic range and a lower halo.