Mid Sweden University

Evaluation of perceived quality of light field images,as well as testing new processing tools, or even assessing the effectiveness of objective quality metrics, relies on the availabilityof test dataset and corresponding quality ratings. This article presents SMART light field image quality dataset. The dataset consists of source images (raw data without optical corrections), compressed images, and annotated subjective quality scores. Furthermore, analysis of perceptual effects of compression on SMART dataset is presented. Next, the impact of image content on the perceived quality is studied with the help of image quality attributes. Finally, the performances of 2D image quality metrics when applied to light field images are analyzed.