安徽大学学报（自然科学版）2025，Vol.49Issue(5)：29-36,8.DOI:10.3969/j.issn.1000-2162.2025.05.004
基于编解码器-生成对抗网络的数字国画文创设计
Digital painting creative design based on encoder-decoder-generative adversarial network
摘要
Abstract
In order to better carry on the inheritance,innovation,and development of traditional culture and enrich the cultural and creative industry,this paper proposed a method of cultural and creative design of digital Chinese painting and landscape painting style transfer based on an encoder-decoder-generative adversarial network.Inspired by the artist's creative process,the framework proposed in this paper included a sketch model consisting of an encoder structure,and a rendering model consisting of a generative adversarial network that combined the sketch with the desired artistic style to generate the final stylistic image by preserving the layout and construction of the painting.Additionally,in the process of style transfer,to ensure that the generated style image was not only consistent with the target style in appearance but also retained the features and structure of the input image in content,this paper also proposed a reconstruction loss function based on content consistency.The experimental results on different landscape painting styles showed that this paper effectively made up for the shortcomings of previous style transfer methods based on Chinese painting.关键词
人工智能/数字文创/编解码器/神经网络Key words
artificial intelligence/digital literature and creation/encoder-decoder/neural network分类
信息技术与安全科学引用本文复制引用
江雪,苏金成,王丽娟..基于编解码器-生成对抗网络的数字国画文创设计[J].安徽大学学报（自然科学版）,2025,49(5):29-36,8.基金项目
国家自然科学基金青年项目(51605204) （51605204）
国家社科基金艺术学一般项目(22BF084) （22BF084）
