Arbitrary Handwriting Image Style Transfer

TL;DR

This paper introduces a neural network based on cGANs for handwriting style transfer, significantly improving the quality and efficiency of imitating handwritten Chinese characters compared to previous methods.

Contribution

The paper presents an improved cGAN model with a new loss function for more effective handwriting style transfer, demonstrating superior results over existing techniques.

Findings

Generated characters are clearer and more realistic.

The method outperforms previous handwriting imitation approaches.

Experimental data confirms high-quality style transfer performance.

Abstract

This paper proposed a method to imitate handwriting style by style transfer. We proposed an neural network model based on conditional generative adversarial networks (cGAN) for handwriting style transfer. This paper improved the loss function on the basis of the GAN. Compared with other handwriting imitation methods, the handwriting style transfer's effect and efficiency have been significantly improved. The experiments showed that the shape of the generated Chinese characters is clear and the analysis of experimental data showed the Generative adversarial networks showed excellent performance in handwriting style transfer. The generated text image is closer to the real handwriting and achieved a better performance in term of handwriting imitation.