Automatic Quality Evaluation of Chinese Character Handwriting by Foreign Students Based on Handwriting Movement Characteristics

The quality of Chinese characters written in hand by foreign students is reflected not only in the result of writing, but also in the writing movement. In this paper, we propose an automatic system to evaluate the quality, training an ensemble of artificial neural networks to classify correct and incorrect handwriting characters through handwriting motion features under the condition of unknown "template words". 25 features are selected from 39 parameters, such as time, space, motion and dynamics. The classification effect of a single classifier is not ideal. The classification accuracy of the ensemble-learning algorithm is about 90%. Writing movement is a very effective representation of the quality of Chinese characters written by foreign students.