Development of a Font Comparison System Using Similarity Metrics

Abstract

Various measures are being used to select the best font for the content. However, the degree of similarity calculated using different measures is different for the same font. In this research, we propose a font comparison system considering different similarity comparison methods by ranking similarities for each font. For measuring similarity, we use MSE, PSNR, SSIM, and MSSSIM for image quality assessment, as well as Euclidean distance and cosine similarity in t-SNE for dimensionality reduction. Relative relations among fonts are obtained by averaging similarity rankings of different similarity comparison methods.