Easy generation of personal Chinese handwritten fonts

Abstract

Creating personal Chinese handwritten font library is a very time-consuming job, with the majority of time spent on users manually writing a large number of Chinese characters. To dramatically cut down the time cost, we propose an efficient solution to generate Chinese handwritten fonts by effectively reusing the sample characters that users write. Our solution first builds a Chinese Character Radical Composition Model based on the images of standard printed characters. The use of contour curve based radical clustering approach facilitates the critical task of learning the model. We then use the model to decide a much smaller set of character that users need to write. The same model is also used to guide the automatic segmentation of user's hand input characters and construction of other characters. Our prototype only needs users to input around 20% characters as usual to create their own qualified handwritten fonts. © 2011 IEEE.