Automatic generation of Chinese calligraphic writings with style imitation

Abstract

A parametric representation of stroke shapes is derived by adopting style imitation, a shape-generation-based process, to compactly represent the shapes of single strokes for the automatic generation of Chinese calligraphic writings. An image-processing-based approach is employed to derive the distance between the two strokes. The concept of stroke context is introduced to determine the shape of a stroke to be produced and the distance between two strokes is defined as the shortest distance between two points. An important difference between personal handwriting and a script generated from a font system is that a human writer writes a certain stroke or character differently each time, while a font system generates the same output. The shape-based criterion used in this study makes direct use of areas, which is more reliable than using shape contours because stroke contour details vary greatly.