Songhua Xu, Francis C. M. Lau, Kwok-Wai Cheung, and Yunhe Pan
We introduce a novel intelligent system which can generate new Chinese calligraphic artwork that meets certain aesthetic requirements automatically. In the machine learning phase, parametric representations of existent calligraphic artwork are derived from input images of calligraphy. Using a six-level hierarchical representation, the acquired knowledge is organized as a small structural stroke database, which is then exploited by a constraint-based analogous reasoning component to create artwork in new styles. A set of geometric constraints is proposed and incorporated into the system for rejecting the aesthetically unacceptable results. The combination of knowledge from various input sources creates a huge space for the intelligent system to explore and produce new calligraphy.