AI-SSIM: Human-Centric Image Assessment through Pseudo-Reference Generation and Logical Consistency Analysis in AI-Generated Visuals
Preprint
- 28 February 2025
- preprint
- Published by Springer Nature in Research Square
Abstract
We present AI-SSIM, a computational image metric for assessing the quality and logical consistency of AI-generated and real-world images. Traditional metrics like structural similarity index measure (SSIM) and multi-scale structural similarity index measure (MS-SSIM) require a ground-truth image, which is often unavailable in AI-generated imagery, and overlook key factors such as logical coherence and content usability. AI-SSIM addresses these gaps by employing advanced pre-trained models to generate a pseudo-reference image, convolution and attention layers to evaluate image quality, and adaptive pooling to minimize distortion during resizing pseudo-reference images. We also designed and statistically validated a multi-item questionnaire for assessing image quality. AI-SSIM was benchmarked against human scales and compared to both full-reference and no-reference metrics, where it demonstrated superior accuracy. The proposed metric has broad applicability, as it can compute scores in both scenarios where ground-truth images are either available or absent.Keywords
All Related Versions
This publication has 81 references indexed in Scilit:
- Controllable image synthesis methods, applications and challenges: a comprehensive surveyArtificial Intelligence Review, 2024
- Survey on video anomaly detection in dynamic scenes with moving camerasArtificial Intelligence Review, 2023
- A survey of the vision transformers and their CNN-transformer based variantsArtificial Intelligence Review, 2023
- A survey of deep learning techniques for machine reading comprehensionArtificial Intelligence Review, 2023
- A Study on Human-Computer Interaction with Text-to/from-Image Game AIs for Diversity Education2023
- High‐fidelity direct contrast synthesis from magnetic resonance fingerprintingMagnetic Resonance in Medicine, 2023
- Compositional Visual Generation with Composable Diffusion Models2022
- Integrating SSIM in GANs to Generate High-Quality Brain MRI Images2021
- GIQA: Generated Image Quality Assessment2020
- PSNR vs SSIM: imperceptibility quality assessment for image steganographyMultimedia Tools and Applications, 2020