品質を考慮したフォント自動作成システム

抄録

In this paper, we propose a font automatic generating system considering quality of a generated font. This system learns a placement of skeleton points in a font to evaluate shape of the font by GRNN (General Regression Neural Network) for each element. Furthermore, using the shape evaluation by the GRNN, the proposed system generates a font considering the quality by revising placement of skeleton points and transformation parameters of the font. In addition, to learn the shape evaluation of the font, a user can evaluate the generated font interactively on the Web. The proposed system has the following three advantages. First, it is possible to evaluate a shape itself of the font. Second, it is possible to make a modification automatically for a skeleton and transformation parameters of a font by shape evaluation. Finally, an interactive shape evaluation of a generated font is possible with low burden.