郭丽娟,任维鑫,魏嵬.基于深度学习的木刻版画图像生成方法[J].中南民族大学学报自然科学版,2022,41(5):592-598
基于深度学习的木刻版画图像生成方法
Method for generating woodcut images based on deep learning
  
DOI:10.12130/znmdzk.20220513
中文关键词: 图像生成  深度学习  神经网络  木刻版画  风格转换
英文关键词: woodcut print  image generation  deep learning  neural network  style transfer
基金项目:陕西省自然科学基金资助项目(2021JM-344);陕西省社科界2020年重大理论与现实问题研究资助项目(2020Z334);陕西省教育厅资助项目(19JK0951);陕西省大学生创新计划训练项目(S202010719100, S202010719056)
作者单位
郭丽娟 延安大学 鲁迅艺术学院延安 716000 
任维鑫 延安大学 物理与电子信息学院延安 716000 
魏嵬 西安理工大学 计算机科学与工程学院西安 710048 
摘要点击次数: 77
全文下载次数: 109
中文摘要:
      木刻版画有着悠久的历史与极高的艺术价值,让机器学习艺术家的木刻版画创作过程与创作风格是一项具有挑战性的研究.目前基于计算机与人工智能的木刻版画图像生成方法多数仅是通过构造硬编码方法去模仿木版画的纹理,色调等模式,缺乏像艺术家一样对创作过程的感知与理解.为了传承木刻版画艺术,利用计算机辅助木刻版画创作, 收集了一个木版画的图像数据集;提出了一种基于深度神经网络的木版画图像生成方法,可将一副摄影图片转化为给定的某种木版画风格的图像;设计了使用文本提示的交互式木版画风格图像的生成算法.实验结果表明:该算法能很好地模拟艺术家的创作风格,可生成各种不同自然场景和人物的木刻版画图像,算法运行高效稳定.
英文摘要:
      Woodcut prints have a long history and high artistic value, and it is a challenging research to make machines learn the process and style of the artist's woodcut prints. Most of the current computer and artificial intelligence-based methods for generating woodcut images are only constructed by hard-coding methods to imitate the texture and tone patterns of woodcuts, lacking the perception and understanding of the creative process as the artists do. In order to pass on the art of woodblock printmaking, a computer-aided woodblock printmaking dataset is collected, a deep neural network-based woodblock image generation method is proposed, which can transform a photographic image into a given woodblock style image, an interactive woodblock style image generation algorithm using textual hints is designed. The experimental results show that the algorithm can simulate the artist's creative style well and can generate woodblock print images of various natural scenes and people.
查看全文   查看/发表评论  下载PDF阅读器
关闭