Research And Application On Artificial Intelligence Of Chinese Calligraphy

As a treasure of Chinese traditional culture,Chinese calligraphy art is loved by many people.Calligraphy lovers learn Chinese calligraphy,not only to practice calligraphy,but also to appreciate the calligraphers' style and connotation.Sometimes calligraphy lovers want to generate a calligraphy plaque in style of some famous calligraphers.Therefore,to generate famous calligraphers' characters with AI is a very meaningful thing.This paper makes use of CADAL to construct a calligraphic resource library,and discusses the limitations of the traditional method of calligraphy character synthesis.We puts forward three novel methods of calligraphy word synthesis,mainly as follows:1.Bayesian relevance feedback based Chinese calligraphy character synthesis.Compared with the traditional method,the method ensures the aesthetic feeling of the generated calligraphic characters,and can effectively synthesize calligraphic characters with specific styles,such as regular script and clerical script.2.Supervised style transfer framework for calligraphy.In order to generate calligraphy characters synthesis in batches,this paper proposes a supervised style transfer framework.The proposed framework generates calligraphy images with high quality and more varied scripts.3.Unsupervised style transfer framework for calligraphy.The proposed framework consists of modified auto-encoders and generative adversarial networks,but with both content images and random vectors as inputs.Experimental results are given,showing that our framework can efficiently synthesize different styles of calligraphy in large quantities without labeling training samples.