Multi-Exposure Image Fusion: Perceptual Quality Assessment Datasets and Methods

This paper presents two datasets for evaluating the perceptual quality of multi-exposure image fusion algorithms.

1. Objective Quality Assessment Dataset (MEF-QE)

The MEF-QE dataset includes 10 scenes, each containing three differently exposed images, a high dynamic range (HDR) reference, and a fused image, totaling 30 images. The images in each scene were standardized to ensure similar brightness, contrast, and color.

2. Subjective Quality Assessment Dataset (MEF-SQE)

The MEF-SQE dataset consists of the same 30 images as the MEF-QE dataset. 20 human observers participated in a subjective quality evaluation, rating each image on a scale of 0-100. The subjective score for each image represents the average rating across all observers.

Both datasets provide valuable resources for evaluating the performance of multi-exposure image fusion algorithms, enabling objective and subjective analysis of the quality of fused images.