Multi-Exposure Image Fusion Algorithm Evaluation: Datasets and Performance Analysis

This paper presents a perceptual evaluation of multi-exposure image fusion algorithms. The research utilizes two datasets:

1. Multi-Exposure Dataset (MED): This dataset contains 142 pairs of images with different exposure levels. Each pair includes 3 to 5 images captured from various cameras and scenes, exhibiting diverse exposure and contrast levels. MED is suitable for evaluating a wide range of image fusion algorithms.

2. Multi-Focus Dataset (MFD): This dataset comprises 45 pairs of multi-focus images. Each pair represents the same scene but with varying focal points. These images capture different objects and textures, allowing for the assessment of multi-focus image fusion algorithm performance.