A proposal project for a blind image quality assessment by learning distortions from the full reference image quality assessments

TL;DR

This paper proposes a new approach for blind image quality assessment that predicts both quality scores and distortion maps without needing the original reference image.

Contribution

It introduces a perspective plan to develop a null reference image quality assessment method capable of estimating distortions and quality scores.

Findings

Conceptual framework for blind quality assessment

Potential to estimate distortion maps without reference images

Foundation for future implementation of null reference IQA

Abstract

This short paper presents a perspective plan to build a null reference image quality assessment. Its main goal is to deliver both the objective score and the distortion map for a given distorted image without the knowledge of its reference image.