Unified Photometric Image Quality (UPIQ)

Unified Photometric Image Quality (UPIQ)数据集是由剑桥大学计算机科学与技术系的研究人员创建的，是目前最大的光度主观图像质量评估数据集。该数据集通过整合四个现有数据集（两个SDR和两个HDR），包含3779张SDR图像和380张HDR图像，总计4159张图像。UPIQ数据集的图像以绝对光度和色度单位表示，对应于显示器发出的光。数据集的质量分数以可解释的JOD单位表示，适用于训练和测试HDR图像质量指标。UPIQ数据集的应用领域包括重新训练现有的HDR指标，展示数据集在亮度感知图像压缩中的效用，并解决因缺乏统一大型数据集而阻碍的机器学习基础HDR图像质量指标的发展问题。

Unified Photometric Image Quality (UPIQ) dataset was created by researchers from the Department of Computer Science and Technology, University of Cambridge, and is currently the largest subjective photometric image quality assessment dataset to date. This dataset integrates four existing datasets (two SDR and two HDR), containing 3779 SDR images and 380 HDR images, with a total of 4159 images. Images in the UPIQ dataset are represented in absolute photometric and chromatic units, corresponding to the light emitted by displays. The quality scores of the UPIQ dataset are expressed in interpretable JOD units, which are suitable for training and testing HDR image quality metrics. The application scenarios of the UPIQ dataset include retraining existing HDR metrics, demonstrating its utility in brightness-aware image compression, and addressing the development challenges of foundational machine learning-based HDR image quality metrics that were previously hindered by the lack of a unified large-scale dataset.