Empirical evaluation of full-reference image quality metrics on MDID database

TL;DR

This paper provides a comprehensive empirical evaluation of 32 full-reference image quality assessment metrics using the MDID database, which includes various distortions applied to pristine images.

Contribution

It offers a detailed comparison of state-of-the-art FR-IQA metrics on a diverse, recently published database with multiple distortion types.

Findings

Identifies the most effective metrics for different distortion types.

Highlights the strengths and weaknesses of current FR-IQA methods.

Provides insights into the correlation between metrics and human perception.

Abstract

In this study, our goal is to give a comprehensive evaluation of 32 state-of-the-art FR-IQA metrics using the recently published MDID. This database contains distorted images derived from a set of reference, pristine images using random types and levels of distortions. Specifically, Gaussian noise, Gaussian blur, contrast change, JPEG noise, and JPEG2000 noise were considered.