Evaluating Non-Reference Image Quality Metrics for AI-Generated Images

Read full text Summarise with VIRAT
Authors: Midhun Kannan B, Dhanalakshmi K
Cited by: 0
References: 16

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

The exponential growth of AI-generated images produced by Generative Adversarial Networks (GANs) and diffusion-based models has created an urgent need for reliable image quality evaluation systems. Traditional full-reference metrics such as PSNR and SSIM require pristine ground-truth images, which are unavailable for generative outputs. This limitation necessitates robust Non- Reference Image Quality Assessment (NR-IQA) frameworks. This paper presents VisionIQ Pro, a hybrid multi- tiered visual quality assessment system that integrates handcrafted statistical signal metrics with deep perceptual Natural Scene Statistics (NSS) metrics such as BRISQUE and NIQE. The system constructs a fused Image Quality Vector and uses a Random Forest regression model trained on KonIQ-10k and LIVE datasets to predict Mean Opinion Score (MOS). Additionally, the framework incorporates Explainable AI through Sobel-based sharpness density mapping and actionable feedback logic. Experimental analysis demonstrates strong correlation (PLCC > 0.85) with human subjective perception. The proposed system serves as a scalable, interpretable, and production-ready solution for automated auditing of AI-generated and real-world images

Fields of Study & Subjects

Image and Video Quality Assessment Digital Media Forensic Detection Image Retrieval and Classification Techniques

Publication details

Journal / source
International Journal of Science Strategic Management and Technology
Volume
02
Issue
03
Pages
1–9
Publication date
26 March 2026
Language
EN