Neural Network Techniques for Image Style Transfer

School of Computer Science, Sichuan Normal University, 610100, Chengdu, China

^* Corresponding author: zhangban@asu.edu.pl

Abstract

With the rapid advancement of deep learning technology, neural networks have achieved remarkable results in the field of image processing. Particularly in image style transfer, neural network-based methods have become a research hotspot. Image style transfer aims to apply the stylistic features of one image to another while preserving its content information, thereby creating images with artistic value. This paper reviews the primary methods of image style transfer, including slow transfer based on image iteration and fast transfer based on model iteration and compared the experimental results of various methods. It further explores the implementation approaches and technical developments of single-style, multi-style, and arbitrary style transfer. Analyzing the strengths and weaknesses of existing methods, this paper examines efficiency improvement, image quality enhancement, and diversity augmentation, highlighting the challenges in content preservation, computational resource requirements, and transfer quality in current style transfer techniques. This paper aims to provide researchers with future research directions.