A Font Multi-style Fusion Model Based on Improved InfoGAN
To solve the problem of complex font structure and difficulty in fusing multi-style features in Chinese characters,a font multi-style fusion method based on improved InfoGAN is proposed.The InfoGAN feature multi-style fusion model has been improved by adjusting the dimension of the input vector and adding a channel attention module.InfoGAN can clarify and regularize style features that are difficult to extract.By improving InfoGAN,multi-style fusion of Chinese font image style features is achieved,and feature vectors that can control Chinese font style are obtained.The improved InfoGAN model is compared with VAE,Beta-VAE and AAE in experiments,and then the effectiveness of channel attention is demonstrated through model ablation experiments.The experimental results show that the model can better separate features of different styles,avoid information overlapping and conflicting,and complete the task of font multi-style fusion effectively and accurately.
integration of multiple features and styleschannel attentionInfoGAN
万方数据
维普
