Case-Based Improvement of Neural Style Transfer for Chinese Painting
Journal of Cases on Information Technology · 07.05.2026 · 28 (1) · 1-18
DOI: 10.4018/jcit.409327 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Cases on Information Technology, erschienen am 07.05.2026. Verfasst von Delong Song und 1 weiteren Autor:innen. Die Arbeit nennt 25 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 3.0).
Abstract (Originalsprache)
Recent neural style transfer advances focus on Western art, neglecting Chinese painting aesthetics like ink gradation, brushwork, and negative space due to convolutional neural network invariance and data scarcity. This study proposes the Ink-Aware Attention Network (IAANet) and a hybrid dataset of 15,000 synthetic paired images and 1,260 real scans across landscapes, floral-bird, and figure paintings, ensuring diversity in ink and content. IAANet uses dual decoders for structure and texture, plus an ink-concentration mask to preserve “bone-method brushwork” and “five tones of ink.” It outperforms adaptive instance normalization, whitening and coloring transform, and style-attentional network (SANet): Mo Gradient Consistency Index reaches 0.78 (0.04 higher than whitening and coloring transform), Fréchet Inception Distance-Ink drops by 23.1 versus adaptive instance normalization, and achieves an expert score of 8.6/10—highest among models. Ablation studies show the dual decoder reduces stroke breaks by 30%, and the ink mask limits color drift below 12%. Deployed efficiently, IAANet runs in 28ms per image (1.4× faster than SANet) with 19.4M parameters (65% of SANet), ideal for digital heritage and art education. This work delivers a case-driven solution bridging artificial intelligence and traditional Chinese art, offering benchmarks for Eastern neural style transfer research.
Zitieren
APA
Song, D., & Liu, X. (2026). Case-Based Improvement of Neural Style Transfer for Chinese Painting. Journal of Cases on Information Technology, 28(1), 1-18. https://doi.org/10.4018/jcit.409327
Vancouver
Song D, Liu X. Case-Based Improvement of Neural Style Transfer for Chinese Painting. Journal of Cases on Information Technology. 2026;28(1):1-18. doi:10.4018/jcit.409327
BibTeX
@article{Song2026,
author = {Song, Delong and Liu, Xiang},
title = {Case-Based Improvement of Neural Style Transfer for Chinese Painting},
journal = {Journal of Cases on Information Technology},
year = {2026},
volume = {28},
number = {1},
pages = {1--18},
doi = {10.4018/jcit.409327}
}
EUVIMED