Learn to Evaluate Image Perceptual Quality Blindly from Statistics of Self-similarity

TL;DR

This paper introduces a novel blind image quality assessment method that leverages statistics of self-similarity within images, inspired by human visual processing, achieving strong prediction accuracy and generalization.

Contribution

The paper proposes a simple yet effective BIQA approach based on self-similarity statistics, enhancing quality prediction without reference images or distortion knowledge.

Findings

Achieves competitive quality prediction performance

Demonstrates strong generalization ability

Outperforms some existing BIQA methods

Abstract

Among the various image quality assessment (IQA) tasks, blind IQA (BIQA) is particularly challenging due to the absence of knowledge about the reference image and distortion type. Features based on natural scene statistics (NSS) have been successfully used in BIQA, while the quality relevance of the feature plays an essential role to the quality prediction performance. Motivated by the fact that the early processing stage in human visual system aims to remove the signal redundancies for efficient visual coding, we propose a simple but very effective BIQA method by computing the statistics of self-similarity (SOS) in an image. Specifically, we calculate the inter-scale similarity and intra-scale similarity of the distorted image, extract the SOS features from these similarities, and learn a regression model to map the SOS features to the subjective quality score. Extensive experiments demonstrate very competitive quality prediction performance and generalization ability of the proposed SOS based BIQA method.