Chinese calligraphy character image synthesis based on retrieval

Abstract

Chinese calligraphy works is a valuable culture heritage written by soft brush. They were scanned into image format, stored in computer in order to enable universal and convenient access. However, the retrieved calligraphy candidates may not satisfy user's expectation due to the size limitation of calligraphy database. This paper propose a way to offer new calligraphy candidates by synthesizing based on the retrieved calligraphy: First, characters contours are extracted to represent calligraphy style. Second, shape matrixes are constructed and peer-to-peer contour points are figured out. Then new calligraphy character candidates that in some degree looks like the retrieved calligraphy characters are generated by employing g-snake algorithm. Finally, the effect of the approach is demonstrated by the implementation. © 2009 Springer-Verlag Berlin Heidelberg.