Neural texture synthesis and style transfer of coal-rock images in coal mine heading faces using very deep convolutional networks
文献类型：期刊论文
| 作者 | Xu, Shuzhan4; Liu, Quansheng4; Yu, Honggan4; Huang, Xing3; Bo, Yin2,4; Lei, Yiming4; Zi, Jiquan1; Yang, Yuanhong1; Zhang, Shoufu1 |
| 刊名 | TUNNELLING AND UNDERGROUND SPACE TECHNOLOGY
 |
| 出版日期 | 2025-03-01 |
| 卷号 | 157页码:20 |
| 关键词 | Coal-rock recognition Neural style transfer Image synthesis Region similarity transform VGG network Bilateral segmentation network |
| ISSN号 | 0886-7798 |
| DOI | 10.1016/j.tust.2024.106342 |
| 英文摘要 | Coal-rock recognition is vital for mining safety and efficiency. Traditional methods are labor-intensive and errorprone, while machine learning and deep learning improve accuracy but are hindered by limited, inconsistent datasets due to challenging mining conditions. To address these issues, this research introduces a novel image synthesis approach leveraging Neural Style Transfer with the VGG-19 model to overcome data scarcity. The style images are derived from 200 real coal face images, while the content images are represented as synthetic grayscale images. Furthermore, the Regional Similarity Transformation Function and the Dragonfly Algorithm are employed to enhance the quality of coal-rock style images. Results from 1,000 iterations indicate that the proposed method substantially improves the quality and diversity of coal-rock images, with interfaces and textural details becoming significantly clearer and more closely resembling the expected coal-rock interfaces seen in the original content images. Additionally, an automated machine learning-based approach is used to generate coal-rock content images, thereby further enhancing the efficiency of the synthesis process. This methodology notably enriches the coal-rock image dataset, bolstering the robustness and accuracy of recognition models and contributing to more efficient mining operations. Synthetic images simulating low-light and dusty environments were created to enhance model robustness. These synthetic images were used to train a coal-rock recognition model, which achieved an impressive 92% accuracy. The findings underscore the effectiveness of synthetic images in overcoming data limitations and strengthening coal-rock recognition systems for real-world applications. |
| 资助项目 | National Natural Science Foun-dation of China[U21A20153] ; Fundamental Research Funds for the Central Universities, China[2042024rs0001] ; Key Research and Development Project of Hubei Province, China[2021BCA133] ; Outstanding Youth Fund Program of Natural Science Foundation of Hubei Province, China[2022CFA084] ; Wuhan Knowledge Innovation Supporting proj-ect, China[2022010801010162] |
| WOS研究方向 | Construction & Building Technology ; Engineering |
| 语种 | 英语 |
| WOS记录号 | WOS:001419692700001 |
| 出版者 | PERGAMON-ELSEVIER SCIENCE LTD |
| 源URL | [http://119.78.100.198/handle/2S6PX9GI/37021]  |
| 专题 | 中科院武汉岩土力学所 |
| 通讯作者 | Liu, Quansheng; Yu, Honggan |
| 作者单位 | 1.Sinohydro Bur 14 Co Ltd, Kunming 650041, Peoples R China 2.Changjiang Inst Survey Planning Design & Res, Wuhan 430010, Peoples R China 3.Chinese Acad Sci, Inst Rock & Soil Mech, State Key Lab Geomech & Geotech Engn, Wuhan 430071, Peoples R China 4.Wuhan Univ, Sch Civil Engn, Wuhan 430072, Peoples R China |
| 推荐引用方式 GB/T 7714 | Xu, Shuzhan,Liu, Quansheng,Yu, Honggan,et al. Neural texture synthesis and style transfer of coal-rock images in coal mine heading faces using very deep convolutional networks[J]. TUNNELLING AND UNDERGROUND SPACE TECHNOLOGY,2025,157:20. |
| APA | Xu, Shuzhan.,Liu, Quansheng.,Yu, Honggan.,Huang, Xing.,Bo, Yin.,...&Zhang, Shoufu.(2025).Neural texture synthesis and style transfer of coal-rock images in coal mine heading faces using very deep convolutional networks.TUNNELLING AND UNDERGROUND SPACE TECHNOLOGY,157,20. |
| MLA | Xu, Shuzhan,et al."Neural texture synthesis and style transfer of coal-rock images in coal mine heading faces using very deep convolutional networks".TUNNELLING AND UNDERGROUND SPACE TECHNOLOGY 157(2025):20. |
入库方式： OAI收割
来源：武汉岩土力学研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。