Artistic Text Style Transfer based on Generative Adversarial Networks

Abstract

Artistic text style transfer refers to transferring the art style of a certain artistic text image to another text image. However, the generation of an artistic word requires professional designers to spend a lot of time to design by hand. The previous style transfer method is to transfer the overall style, which is not suitable for highly structured text images. In this paper, we propose a kind of artistic text style transfer based on Generative Adversarial Networks. By disassociating the font features of the text image and the style features of the artistic text image, and then reorganizing the font features and style features, a new artistic text image is generated. Experimental results show that the method in this paper can generate better style transfer effect than other methods.