基于注意力机制的高光谱图像降维在纸质文物霉斑识别的研究

doi:10.3964/j.issn.1000-0593(2025)01-0246-10

摘要
参考文献
相关文章 (15)

全文: PDF (44291 KB)
输出: BibTeX | EndNote (RIS)

摘要：纸质文物作为文物传承的重要工具，用于记录不同时期人类历史及人文风貌，其在保存过程中极易受到霉菌等微生物的侵害。霉菌会加速纤维素的降解，在纸张表面生成霉斑，并且散落的孢子会随空气流动大范围传播，增加其他纸质文物发生霉变的风险。因此，定期对纸质文物进行霉斑检测对了解纸质文物现状和纸质文物修复至关重要。高光谱成像技术是一种非接触性、非破坏性的检测技术，能同时获得空间数据和光谱数据，与计算机技术结合可以实现纸质文物的大批次实时无损检测。针对黑曲霉这一广泛出现的霉菌，提出一种基于注意力机制的高光谱数据降维方法，通过采集其高光谱数据，实现了高光谱冗余数据的自适应预处理。采集了来自重庆中国三峡博物馆提供的20份纸质文物黑曲霉霉斑样本，使用ENVI软件分析得出在413~855 nm波段范围内，黑曲霉霉斑感染区域和健康区域的平均光谱曲线，平均反射率差异明显；在855~1 021 nm波段范围内，黑曲霉霉斑感染区域和墨迹区域的平均光谱曲线，平均反射率差异明显。文中将所提出方法与传统主成分分析和独立成分分析预处理方法分别处理原始高光谱数据，并将结果在经典U-Net、SegNet、DeepLabV3+和PSPNet四个语义分割网络上进行了对比。结果表明，该算法预处理的数据在U-Net和SegNet经典网络中有明显优势，相较于主成分分析法和独立成分分析法，霉斑识别精度取得了较大提升达到89.49%和88.46%，验证了本文所提出算法的有效性，为文物保护领域提供有效的支撑和新的思路。

关键词：高光谱数据预处理；霉斑识别；纸质文物；注意力机制；图像分割

Abstract：Paper cultural relics are important for heritage transmission as they record human history and humanities in different periods. However, they are highly susceptible to microorganisms such as mold during preservation. Mold can accelerate the degradation of cellulose, generating mold on the surface of paper. Scattered spores can spread widely with airflow, increasing the risk of mold on other paper cultural relics. Regular mold spot detection is crucial for understanding paper artifacts' status and restoration. Hyperspectral imaging technology is a non-contact and non-destructive detection method that simultaneously obtains spatial and spectral data. This technology can be combined with computer technology to enable large batches of real-time, non-destructive testing of paper cultural relics. This paper proposes a method for reducing the dimensionality of hyperspectral data for Aspergillus niger, a commonly occurring mold. The method is based on the attention mechanism and allows for adaptive preprocessing of hyperspectral redundant data. This paper reports on the collection of 20 samples of Aspergillus niger, mold spots on paper artifacts provided by the Chongqing China Three Gorges Museum. The average spectral curves of the infected and healthy areas are analyzed using ENVI software in the 413~855 and 855~1 021nm bands. The results showed a significant difference in average reflectance between the two areas. The paper compares the proposed method with traditional principal component analysis and independent component analysis preprocessing methods for processing original hyperspectral data. The results are then experimented on four semantic segmentation networks: classical U-Net, SegNet, DeepLabV3+, and PSPNet. The experimental results demonstrate that the preprocessed data produced by the algorithm presented in this paper exhibit significant advantages over the classical U-Net and SegNet networks. Furthermore, compared to the principal component analysis method and independent component analysis method, the accuracy of mold spot identification has improved significantly by 89.49% and 88.46%, respectively. These results confirm the effectiveness of the proposed algorithm and provide valuable support and new ideas for the field of cultural relics protection.

Key words：Hyperspectral data preprocessing; Mole spots detection; Paper artifacts; Attention mechanism; Image segmentation

收稿日期: 2023-11-21 修订日期: 2024-04-24

中图分类号:

TN29

基金资助: 国家重点研发计划项目（2020YFC1522500）, 重庆市教委科学技术研究项目（KJQN202201110），重庆市科委技术创新与应用发展专项一般项目（cstc2020jscx-msxmX0097）, 重庆市高校创新研究项目(CXQT21035), 重庆市研究生科研创新项目（CYS23660）资助

通讯作者: 唐欢，王建旭 E-mail: wangjianxu@cqut.edu.cn; tanghuan3gm@163.com

作者简介: 汤斌，1985年生，重庆理工大学副教授 e-mail: tangbin@cqut.edu.cn

引用本文:

汤斌，贺渝龙，唐欢，龙邹荣，王建旭，谭博文，覃丹，罗希玲，赵明富. 基于注意力机制的高光谱图像降维在纸质文物霉斑识别的研究[J]. 光谱学与光谱分析, 2025, 45(01): 246-255.
TANG Bin, HE Yu-long, TANG Huan, LONG Zou-rong, WANG Jian-xu, TAN Bo-wen, QIN Dan, LUO Xi-ling, ZHAO Ming-fu. Attention Mechanism Based Hyperspectral Image Dimensionality Reduction for Mold Spot Recognition in Paper Artifacts. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2025, 45(01): 246-255.

链接本文:

https://www.gpxygpfx.com/CN/10.3964/j.issn.1000-0593(2025)01-0246-10 或 https://www.gpxygpfx.com/CN/Y2025/V45/I01/246

[1] Cortesão M, de Haas A, Unterbusch R, et al. Frontiers in Microbiology, 2020, 11: 560.
[2] Caselli E, Pancaldi S, Baldisserotto C, et al. PLoS One, 2018, 13(12): e0207630.
[3] CHEN Jia-li, LIU Xing-xing, HAN Qiu-lin, et al(陈佳丽, 刘星星, 韩秋琳, 等). Cultural Relics World(文物天地), 2023，(12): 81.
[4] Boniek D, Bonadio L, Damaceno Q S, et al. Canadian Journal of Microbiology, 2020, 66(10): 586.
[5] ZHANG Nuo, XU Sen(张诺, 徐森). Sciences of Conservation and Archaeology(文物保护与考古科学), 2020, 32(1): 77.
[6] ZHANG Nuo, CHEN Xiao-li, DING Li-ping, et al(张诺, 陈潇俐, 丁丽平, 等). Journal of Nanjing University of Technology(Natural Science Edition)[南京工业大学学报(自然科学版)], 2022, 44(4): 450.
[7] Apacionado B V, Ahamed T. Sensors, 2023, 23(20): 8519.
[8] Jung D H, Kim J D, Kim H Y, et al. Frontiers in Plant Science, 2022, 13: 837020.
[9] Yuan D, Jiang J, Qi X, et al. Infrared Physics & Technology, 2020, 111: 103518.
[10] Cong S, Sun J, Mao H, et al. Journal of the Science of Food and Agriculture, 2018, 98(4): 1453.
[11] Treepong P, Theera-Ampornpunt N. Current Research in Food Science, 2023, 7: 100574.
[12] Solórzano J V, Mas J F, Gao Y, et al. Remote Sensing, 2021, 13(18): 3600.
[13] Wang J, Liu W, Gou A. Urban Forestry & Urban Greening, 2022, 69: 127488.
[14] Liu Y, Bai X, Wang J, et al. Engineering Applications of Artificial Intelligence, 2024, 127: 107260.
[15] XIONG Bin, ZHANG Shuang-de(熊彬, 张双德). Remote Sensing Information(遥感信息), 2023, 38(4): 73.
[16] Xie W, Lei J, Yang J, et al. IEEE Transactions on Geoscience and Remote Sensing, 2019, 58(3): 2015.
[17] Jiao C, Chen C, McGarvey R G, et al. ISPRS Journal of Photogrammetry and Remote Sensing, 2018, 146: 235.
[18] Greenacre M, Groenen P J F, Hastie T, et al. Nature Reviews Methods Primers, 2023, 3: 22.
[19] LIN Yu-kun, WANG Nan, ZHANG Li-fu, et al(林昱坤, 王楠, 张立福, 等). National Remote Sensing Bulletin(遥感学报), 2019, 23(6): 1167.
[20] Guo M H, Xu T X, Liu J J, et al. Computational Visual Media, 2022, 8(3): 331.