1. Key Laboratory of 3D Information Acquisition and Application, Ministry of Education, Capital Normal University, Beijing 100048, China
2. Engineering Research Center of Spatial Information Technology, Ministry of Education, Capital Normal University, Beijing 100048, China
3. Chinese Academy of Cultural Heritage, Beijing 100029, China
Abstract：Ancient murals have often suffered from blurring and loss of line structures over time, making their interpretation challenging. Hyperspectral imaging technology, capable of capturing subtle variations in material and energy, provides valuable information for enhancing these faint or missing details. Therefore, this paper proposes a method for enhancing mural line information that combines the wavelet transform with hyperspectral imaging data. Firstly, the dimensionality of the mural hyperspectral image was reduced by using the Minimum Noise Fraction (MNF) transform. The optimal MNF band image was selected using the maximum average gradient method. The MNF results were used to extract the pure end members, and the corresponding abundance maps were inverted by fully constrained least squares spectral unmixing. The line abundance map was combined with the optimal MNF band image through band operations to obtain a linear feature-enhanced image. Then, the true color image was synthesized using the inverse MNF transformation. Both the linear feature-enhanced image and the true color image were processed with Gaussian filtering to enhance detail. Haar wavelet decomposition was applied to both images. The corresponding high-frequency components were fused, while the low-frequency component from the true color image was retained to reconstruct the final color image with enhanced line features. The experimental validation on murals from Yiju Temple (Shanxi) shows that, in comparison with a PCA-based enhancement technique, the proposed approach achieves an increase of 0.083 7 in average gradient and 15.253 1 in edge intensity, indicating more effective enhancement of line features and offering valuable insights for mural preservation and restoration efforts.
[1] Capobianco G, Prestileo F, Serranti S, et al. Periodico di Mineralogia, 2015, 84(3A): 407.
[2] Liang H. Applied Physics A, 2012, 106(2): 309.
[3] Tonazzini A, Salerno E, Abdel-Salam Z A, et al. Journal of Advanced Research, 2019, 17: 31.
[4] Pillay R, Hardeberg J Y, George S. Journal of the American Institute for Conservation, 2019, 58(1-2): 3.
[5] Delaney J K, Zeibel J G, Thoury M, et al. Applied Spectroscopy, 2010, 64(6): 584.
[6] LI Jun-feng, WAN Xiao-xia(李俊锋, 万晓霞). Spectroscopy and Spectral Analysis(光谱学与光谱分析), 2018, 38(1): 200.
[7] Cao J, Li Y, Cui H, et al. Heritage Science, 2018, 6: 67.
[8] Deng X, Yu Y. Heritage Science, 2022, 10: 163.
[9] Wu M, Jia M, Wang J. Applied Sciences, 2022, 12(21): 10940.
[10] Yu K, Li Y, Yan J, et al. Proceedings of SPIE, Optics for Arts, Architecture, and Archaeology VIII. 2021, 11784: 1178409.
[11] Hou M, Zhou P, Lv S, et al. Journal of Cultural Heritage, 2018, 34: 136.
[12] Zhou P, Hou M, Lv S, et al. Remote Sensing, 2019, 11(11): 1384.
[13] Pan N, Hou M, Lv S, et al. Journal of Cultural Heritage, 2017, 27: 80.
[14] HOU Miao-le, LEI Yong, LU Xin, et al(侯妙乐, 雷 勇, 芦 鑫, 等). Science of Surveying and Mapping(测绘科学), 2014, 39(10): 89.
[15] MAO Jin-cheng, LÜ Shu-qiang, HOU Miao-le, et al(毛锦程, 吕书强, 侯妙乐, 等). Journal of Graphics(图学学报), 2022, 43(3): 425.
[16] CAO Peng-hui, LÜ Shu-qiang, HOU Miao-le, et al(曹鹏辉, 吕书强, 侯妙乐, 等). Sciences of Conservation and Archaeology(文物保护与考古科学), 2021, 33(1): 26.
[17] Wu T, Cheng Q, Wang J, et al. Archaeological and Anthropological Sciences, 2019, 11(8): 4147.
[18] Pan N, Hou M. The Extraction and Fusion of Faint Mural Based on Feature Transform of Hyperspectral Images. 2016 4th International Workshop on Earth Observation and Remote Sensing Applications (EORSA), 2016, 161.
[19] Peng J, Yu K, Wang J, et al. Journal of Cultural Heritage, 2019, 36: 32.
[20] Han D, Ma L, Ma S, et al. Journal of Physics: Conference Series, 2019, 1237(3): 032028.
[21] Han X, Hou M, Zhu G, et al. Tehnički Vjesnik—Technical Gazette, 2015, 22 (6):1567.
[22] LI Jing-yu, ZHANG Rong-fen, LIU Yu-hong(李景玉, 张荣芬, 刘宇红). Optical Technique(光学技术), 2021, 47(2): 217.
[23] XU Jing-xiu, ZHANG Qing(徐景秀, 张 青). Computer Engineering & Science(计算机工程与科学), 2022, 44(1): 92.
