Font Style Conversion Based on Deep Learning

In view of the cost of traditional design of new fonts, a method of combining deep learning for font style conversion is proposed. By using a U-Net type network structure combining the training method of generative adversarial network, supervised learning part of fonts, the font style conversion ability of the font generator network is constantly enhanced so that the fonts can be converted to another style through the generator network. The experimental results show that this method convert font structure generated clear and smooth, less noise, and the original character of the same font highly consistent in size, weight, style etc.