Reconstructing dynamic font-based Chinese characters using support vector machine
Hiroyuki Fujioka, Wenli Zhu, Akinori Hidaka, Hiroyuki Kano · 2017
This paper considers a problem of reconstructing a typeface of Chinese characters. First, Chinese characters are generated by using the so-called dynamic font method as a basic tool. Then, we develop a scheme for reconstructing characters to cursive characters with natural "Renmen" which can depict continuity of the adjacent strokes. For realizing the natural Renmen, we propose a method for judging stroke connectivity by introducing kernel support vector machine when we reconstruct the character. It is shown that the reconstruction problem is formulated as a smoothing spline problem with constraints expressed in terms of stroke connectivity. The performances are demonstrated by some experimental studies.