Research on Artistic Font Multi-style Fusion Technology Based on Conditional Normalization Network
In order to solve the problems of style uniformity,style overflow and residual black dots in multi-style font fusion,a method of multi-style font image fusion based on conditional normalization network is proposed.This method employs a font multi-style fusion model that adjusts the output of the pre-trained VGG19Net based on different style parameters.By extracting features from content images and multiple style font images separately,and using a new style loss function to measure the differences between content images and style font images,multiple style features are blended to generate new artistic font style images.The key components of this method are the convolutional layers and conditional normalization layers,which enable the transfer and fusion of multiple styles.Experimental results show that,compared to other two methods,the proposed method achieves an average improvement of 1.2%in similarity of content and 2.7%in similarity of style.
万方数据
维普
