Multi-exposure Image Fusion via Feature Distribution Weights
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multi-exposure image fusion (MEF) methods fail to consider the overall feature distribution of images, leading to suboptimal weight determination and resulting in low fusion quality, complex computation, and incomplete information in fused images.
Innovation Solution
The proposed MEF method performs color space transformation, determines luminance, exposure, and local gradient weights, and combines them using Gaussian kernel functions to adaptively compute pixel weights, enabling high-quality fusion with simple computation and abundant information inclusion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If existing pixel-wise MEF method only designs weight function according to feature distribution of input image, then the computation is simple, but the fusion quality is low and parameters need to be set manually
Solution Approach 1:
The patent transforms the weight determination from static parameter-based methods to dynamic feature distribution-based methods. By using luminance distribution, exposure distribution, and local gradient characteristics as adaptive parameters, the system automatically adjusts weights without manual parameter setting, thereby improving fusion quality while maintaining computational efficiency.
Solution Approach 2:
The weight function determines its own parameters by analyzing the feature distribution of the input images themselves. The luminance distribution weight, exposure distribution weight, and local gradient weight are all derived from the image data without requiring external parameter input, enabling the system to adapt automatically to different input conditions.
2Measurement precision
If patch-wise approach is used to divide image into patches and determine weight map, then the weight determination considers local features, but spatial artifacts appear between patches and additional processing is required
Solution Approach 1:
The patent segments the weight determination into three distinct feature dimensions: luminance distribution, exposure distribution, and local gradient. Each dimension contributes to the final weight map, allowing comprehensive local feature consideration while maintaining a unified pixel-wise framework that avoids patch boundary artifacts.
Solution Approach 2:
The patent merges multiple feature distributions (luminance, exposure, gradient) into a single comprehensive weight map through multiplication and normalization. This unified approach integrates local feature information from different sources without requiring separate patch-based processing, thereby avoiding spatial artifacts while maintaining processing efficiency.
3Loss of information
If multiple LDR images with different exposures are input to determine weight maps, then the HDR scene information can be captured, but the computation becomes complicated and time-consuming
Solution Approach 1:
The patent extracts only the essential feature distributions (luminance, exposure, gradient) from the multi-exposure LDR images that are necessary for weight determination. By focusing on these key features rather than processing all image data equally, the method captures HDR scene information efficiently while reducing computational burden.
Solution Approach 2:
The patent performs preliminary feature analysis by computing luminance distribution, exposure distribution, and local gradient characteristics before the actual weight map generation. These pre-computed feature distributions are then directly used in the weight function, avoiding redundant computations and accelerating the overall processing time.
Data Source
AI summary
The present disclosure provides a multi-exposure image fusion (MEF) method based on a feature distribution weight of a multi-exposure image, including: performing color space transformation (CST) on an image, determining a luminance distribution weight of the image, determining an exposure distribution weight of the image, determining a local gradient weight of the image, determining a final weight, and determining a fused image. The present disclosure combines the luminance distribution weight of the image, the exposure distribution weight of the image and the local gradient weight of the image to obtain the final weight, and fuses the input image and the weight with the existing pyramid-based multi-resolution fusion method to obtain the fused image, thereby solving the technical problem that an existing MEF method does not consider the overall feature distribution of the multi-exposure image.