Semantic Usability of Digital Images -Ontology-Based Readability Assessment

This paper introduces an ontology-driven method for assessing the semantic usability of digitized images, with a focus on readability and layout readiness, emphasizing practical metrics relevant for text extraction and annotation workflows: Optical Character Recognition (OCR) performance, line segmentation quality, script and annotation compatibility. While previous work has addressed blind Image Quality Assessment (IQA) using perceptual and statistical methods, such approaches often overlook a critical dimension - the semantic usability of images. The method assesses digitized images using a combination of objective metrics (e.g., resolution, noise, compression) and semantic tests (e.g. OCR accuracy, layout analysis) with future goal of implementing metadata completeness, Linked Open Data and International Image Interoperability Framework (LOD/IIIF) compatibility in order to provide a composite usability score. The proposed model is grounded in the Image Authenticity and Quality Ontology (IAQO), a lightweight ontology designed to represent assessment outputs in a standardized, machine-readable format. A dataset of 800 public-domain manuscript images was analyzed, yielding average semantic usability scores ranging from 73.8 to 85.4 (out of 100) across repositories. Results demonstrate that high visual fidelity does not always imply high semantic usability but also the framework’s potential to improve digitizing practices. This approach supports FAIR (Findable, Accessible, Interoperable, Reusable)-compliant image assessment.