Research and Application of Image Style Transfer Method

Abstract

Image style transfer is a technology that uses specific algorithms in deep learning to separate image content and style. It combines the style of one image with the content of another. This technology can be applied to film and television special effects, artistic creation and other fields. It is a research hotspot in the field of image processing in recent years. This paper summarizes the scholars' research on image style transfer technology. Convolution neural network is used to realize image style transfer. The style image and content image are input into the convolution neural network. The loss functions of content image and style image are defined. Vgg-19 model is established, and the loss function is weighted into a complete loss function. In each iteration, the pixel value of the image is adjusted according to the total loss function. After many iterations, the fusion effect image of content and style is obtained.