With the rise of new media, various forms of communication have emerged, and as a result, online communication has diversified. In order to improve the efficiency and competitiveness of their contents, people’s interest in typeface design is rapidly...
With the rise of new media, various forms of communication have emerged, and as a result, online communication has diversified. In order to improve the efficiency and competitiveness of their contents, people’s interest in typeface design is rapidly increasing, and the number of fonts and the designs are diversifying.
Although the influence of typefaces is rapidly increasing, the Hangul font system stores font information only by font names or font manufacturing company names. There are limits when trying to identify the fonts with just the shape of the characters. If the font used is not saved in the computer, the user needs to replace the font with a font similar to the one used in the file, so finding the font can be difficult with the current system since checking all the fonts one by one is time consuming. Also, if the users design a font, they need to find if there are existing ones that are similar to the one they designed which is an exhausting job currently. It is practically impossible to identify and use thousands of Hangul fonts.
Therefore, the influence and importance of typefaces has increased rapidly, but the lack of the Hangul font system has caused inconvenience for both users of fonts and designers who create fonts, making it difficult to use them efficiently. Recently, an increasing number of attempts have been made to solve the typeface-related problem using deep learning technology, but Hangul is a combination of characters, and unlike Roman characters, its structure is complicated, making it difficult to analyze the characters.
The structure of Hangul consists of Stroke Element, Skeleton, and Spacing. Therefore, we believe that analyzing fonts according to the characteristics of Hangul will produce better results.
In this study, we aim to analyze the characteristics of Hangul fonts based on the information from Stroke Elements of Hangul fonts which is different from the previous attempts to analyze by looking at the font as a whole. Furthermore, we propose a method to automatically recommend similar fonts based on the information.