Image Arbitrary Style Transfer via Self-Attention Mechanism Based on Feature Fusion

Arbitrary style transfer is a technique for transferring an ordinary photograph to an image with another artistic style, which not only provides rich materials for film and game production, but also can be used for image restoration. To solve the problems that stylized images cannot maintain spatial consistency and content structure distortion, this paper proposes an arbitrary style transfer via self-attention network method. This algorithm can match the closest semantic style feature to the content feature, and adjust the local style efficiently and flexibly according to the semantic spatial distribution of the content image. Moreover, it can retain more original information about the structure. The experimental results show that the method can transfer the image into different styles with higher quality and better visual effects.