Image Quality Assessment: Investigating Causal Perceptual Effects with Abductive Counterfactual Inference

TL;DR

This paper introduces a novel full-reference image quality assessment method that uses abductive counterfactual inference to uncover causal relationships between deep features and human perceptual responses, improving robustness and interpretability.

Contribution

The paper presents a causal inference-based IQA model that is architecture-independent and enhances understanding of perceptual distortions compared to existing methods.

Findings

Achieves superior correlation with human judgments across benchmarks.

Demonstrates robustness across various deep network backbones.

Provides interpretable quality scores grounded in causal reasoning.

Abstract

Existing full-reference image quality assessment (FR-IQA) methods often fail to capture the complex causal mechanisms that underlie human perceptual responses to image distortions, limiting their ability to generalize across diverse scenarios. In this paper, we propose an FR-IQA method based on abductive counterfactual inference to investigate the causal relationships between deep network features and perceptual distortions. First, we explore the causal effects of deep features on perception and integrate causal reasoning with feature comparison, constructing a model that effectively handles complex distortion types across different IQA scenarios. Second, the analysis of the perceptual causal correlations of our proposed method is independent of the backbone architecture and thus can be applied to a variety of deep networks. Through abductive counterfactual experiments, we validate the proposed causal relationships, confirming the model's superior perceptual relevance and interpretability of quality scores. The experimental results demonstrate the robustness and effectiveness of the method, providing competitive quality predictions across multiple benchmarks. The source code is available at https://anonymous.4open.science/r/DeepCausalQuality-25BC.