@article{xu_fast_2025,
	title = {A fast multi-scale textile pattern generation method combining layered loss and convolutional attention},
	volume = {76},
	issn = {12225347},
	url = {https://www.revistaindustriatextila.ro/images/2025/3/11%20GUODONG%20XU%20INDUSTRIA%20TEXTILA%20no.3_2025.pdf},
	doi = {10.35530/IT.076.03.2024126},
	abstract = {This paper introduces a neural network-based algorithm for the rapid multi-scale synthesis of textile patterns. The
algorithm achieves comprehensive textile pattern style reconstruction by utilising low-level feature loss, represented by
the Gram matrix, to capture colour and texture, and high-level feature loss, represented by the Wasserstein distance, to
capture complex structures and semantic content. The convolutional attention feature enhancement module is
incorporated to improve pattern detail clarity and overall visual quality by emphasising significant features and
suppressing irrelevant information. Furthermore, the multi-scale optimisation module enhances texture and layering by
optimising the image at different scales. Experimental results demonstrate that, compared to existing methods, this
approach offers superior visual effects in pattern synthesis and scalability in pattern size. It not only generates
high-quality textile patterns but also excels in managing complex textures and maintaining semantic consistency. This
method aids designers in extending pattern designs and advances the intelligent development of textile design and
production.},
	number = {03},
	urldate = {2025-07-08},
	journal = {Industria Textila},
	author = {Xu, Guodong and Chen, Yu and Hua, Jian},
	month = jun,
	year = {2025},
	pages = {397--406},
}