Abstract

Human eyesight has deteriorated over the past decades. One of the main reasons is that we have to read a lot. The long-term goal of our research is to identify legible print-fonts or invent some new print-fonts which will sooth our eyes and prevent damages from too much reading. In this thesis, from a pattern recognition point of view, we focus on the study of the effect of font design characteristics on font legibility. First, the effects of typeface on legibility are investigated in an experiment. Studied are nine sans serif typefaces and nine serif typefaces. OCR systems, one of the most successful applications of machine reading, were used in legibility measurements. The results show how the legibility is affected by different fonts. Second, the detailed font design characteristics were studied. They include serif and sans serif design, x-height, ascender and descender of the typeface, the inter-letter spacing setting, the weight of the stroke and the distinctive character features. In order to measure them mathematically and objectively, each characteristic is represented by one or two legibility factors and the extraction methods of some factors are proposed. In addition, the statistics on the common recognition errors of the eighteen fonts are studied. How the legibility factors affect and contribute to legibility is analyzed. Moreover, design suggestions are made on typeface characteristics to maximize typeface legibility. Finally, one of the important font characteristics, inter-letter spacing, is studied. We addressed the issue of how the condensed or expanded inter-letter spacing setting contributes to legibility. By measuring the legibility of different inter-letter spacing settings for a given typeface, and analyzing their corresponding touching errors, better inter-letter spacings for reading are suggested