基于深度学习的动漫风格迁移研究综述  

A Summary of Research on Animation Style Based on Deep Learning

作　　者：聂雄锋 王俊英[1] 江曙 陈晗晗 NIE Xiongfeng;WANG Junying;JIANG Shu;CHEN Hanhan(College of Computer and Information,Three Gorges University,Yichang 443002,China)

机构地区：[1]三峡大学计算机与信息学院,湖北宜昌443002

出　　处：《长江信息通信》2022年第2期37-39,共3页Changjiang Information & Communications

摘　　要：图像动漫化技术的发展对我国动漫产业影响巨大。目前基于深度学习的动漫风格迁移研究是一项热门的研究方向,相关算法层出不穷。文章对动漫风格迁移领域现有的主流方法和代表性工作进行了归纳和讨论,分析了该领域所使用的主要深度神经网络模型,并按照动漫风格迁移方法所解决的不同实际问题,将其归纳为风景动漫迁移、人像动漫迁移和视频帧动漫迁移三类并对每个类别进行了分析和讨论,最后总结了基于深度学习的动漫风格迁移目前存在的问题和未来研究方向。The development of image animation technology is huge for my country’s anime industry.At present,the research based on deep learning is a popular research direction,and the correlation algorithm is endless.The article summarizes and discusses the existing mainstream methods and representative work in the field of anime style migration.It analyzes the main depth neural network model used in this area,and summarizes it according to different practical problems solved by anime style migration method.Three categories of scenic animation migration,portrait animation migration,and video frame anime migration,analyzing and discussing each category,and finally summarizes the current problems and future research directions based on deep learning anime migration.

关 键 词：深度学习 动漫风格迁移 生成对抗网络 迁移学习 卷积神经网络 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]