TY - GEN
T1 - A Structure-Aware Deep Learning Network for the Transfer of Chinese Landscape Painting Style
AU - Way, Der Lor
AU - Lo, Chang Hao
AU - Wei, Yu Hsien
AU - Shih, Zen Chung
N1 - Publisher Copyright:
© 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.
PY - 2023
Y1 - 2023
N2 - Recently, deep learning technology has made a breakthrough in computer vision, image processing, and other fields. Some researchers suggested neural style transfer method using a convolutional neural network (CNN). They established the correlation of features in a neural network to be treated as the style. However, their performance is unacceptable for Chinese landscape painting. According to the property of the Chinese landscape painting, this paper proposes a novel two stage style transfer method that imitates multiple styles of Chinese landscape painting based on deep learning. The structure of an input photo was simplified in the first stage. Then, the result of the first stage was transferred into the final stylized image in second stage. A generative adversarial network (GAN) is applied to train in each stage. Besides, a novel loss function was proposed to keep the shape of the content image. Finally, our method haves successfully imitated several styles of Chinese Landscape ink painting.
AB - Recently, deep learning technology has made a breakthrough in computer vision, image processing, and other fields. Some researchers suggested neural style transfer method using a convolutional neural network (CNN). They established the correlation of features in a neural network to be treated as the style. However, their performance is unacceptable for Chinese landscape painting. According to the property of the Chinese landscape painting, this paper proposes a novel two stage style transfer method that imitates multiple styles of Chinese landscape painting based on deep learning. The structure of an input photo was simplified in the first stage. Then, the result of the first stage was transferred into the final stylized image in second stage. A generative adversarial network (GAN) is applied to train in each stage. Besides, a novel loss function was proposed to keep the shape of the content image. Finally, our method haves successfully imitated several styles of Chinese Landscape ink painting.
KW - Chinese landscape painting
KW - generative adversarial network (GAN)
KW - loss function
KW - style transfer
UR - http://www.scopus.com/inward/record.url?scp=85169040609&partnerID=8YFLogxK
U2 - 10.1007/978-3-031-34732-0_25
DO - 10.1007/978-3-031-34732-0_25
M3 - Conference contribution
AN - SCOPUS:85169040609
SN - 9783031347313
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 326
EP - 337
BT - Culture and Computing - 11th International Conference, C and C 2023, Held as Part of the 25th HCI International Conference, HCII 2023, Proceedings
A2 - Rauterberg, Matthias
PB - Springer Science and Business Media Deutschland GmbH
T2 - 11th International Conference on Culture and Computing, C and C 2023, held as part of the 25th International Conference on Human-Computer Interaction, HCII 2023
Y2 - 23 July 2023 through 28 July 2023
ER -