In this paper, a hierarchical architecture to synthesizing Chinese handwritten character models is proposed for e-learning. In this approach, each handwritten stroke is represented by its skeleton, which is further described by several segments of 3D cubic Bezier curves. Due to the compositional nature, not only can the synthesized Chinese character models be displayed by stoke order, but also they can further be decomposed to show their internal structures, which are very important in Chinese handwriting instruction. Besides the above two advantages, the synthesized character models also have the following important merits: (1) similar to original samples, (2) smoother than original ones, (3) scalable to any size without jagging, and (4) storage efficiency.
第四屆AEARU網路教學國際學術研討會論文集=Proceedings of the Fourth AEARU Workshop on Network Education (AEARU 2006)，6頁