摘要
目的 针对图像智能生成中图像风格的多元性问题,探索能够生成特定风格图像的智能生成方法。方法 以抽象和具象图像结合为研究对象,通过计算机将两种风格迁移,重构出保留特定风格的图像,得到其中捕捉画面特定风格的图像。结果 生成的图像质量较高,具备抽象图案和具象图案相互结合的特点。结论 生成对抗网络对绘画艺术作品的特定风格的学习,用风格迁移智能生成图像,取得了较好效果,能减少创意图像制作的重复性操作,提高图像利用的艺术性和丰富性,为智能赋能图像的再创造提供了新思路。
The work aims to explore the intelligent generation method that can generate images of specific styles in view of the problem of plurality of image styles in intelligent image generation.Taking abstract and figurative images as the research object,the computer was used to transfer the two styles,reconstruct the image retaining a specific style,and obtain an image that captured a specific style of the picture.The resulting image was in high quality and had the charac-teristics of both abstract pattern and figurative image.For learning the specific style of painting artwork by generative adversarial network and intelligently generating images with style transfer,good results have been achieved,which can reduce the repetitive operation of creative image production,improve the artistry and richness of image utilization,and provide a new idea for the re-creation of intelligently empowered images.
作者
徐术珂
吴倩
XU Shuke;WU Qian(Southwest Minzu University,Chengdu 610000,China;Sichuan Technology and Business University,Chengdu 610000,China)
出处
《包装工程》
CAS
北大核心
2024年第2期365-373,共9页
Packaging Engineering
基金
中央高校课题(2023SYB29)。
关键词
风格迁移算法
艺术
文创产品
style transfer algorithm
artistic
cultural and creative product
