Classifying Digital Arabic Fonts Based on Design Characteristics

Abstract

PANOSE is a system for classifying and matching typefaces based on their design features. It composes of 10 digits where each digit represents a specific design feature. It uses in several font management tools as option for ordering and searching fonts regarding their design features. Also, it uses in font replacement process when an application or an operating system detects a missing font in an immigrant document or website. Currently, PANOSE is only defined for fonts that have Latin characters. In this paper, a new model is defined to extend PANOSE coverage to include Arabic characters. The reliability of the model has been evaluated by conducting two clustering processes for 30 fonts of Naskh style. Our model, PANOSE-A, is used to measure a similarity matrix for one clustering process while the other process used a similarity matrix produced by using font matching tool. The Silhouette coefficient for both clustering was calculated, giving 97.6 with the font matching tool and 96.6 with PANOSE-A, which indicate the validity and consistency of the results.