受保存条件影响, 很多纸质文物表面会形成狐斑(foxing), 如果不能进行有效监测和科学判断, 会进而影响纸质文物安全。 纸质文物狐斑病害检测存在滞后性、 主观性等问题, 对于书画藏品被墨色、 颜料及印章等覆盖的区域更是难以通过肉眼进行识别, 因此, 基于文物的预防性保护理念, 亟待开发对于狐斑高效、 精确识别的无损检测技术。 可见光-近红外高光谱图像结合了光谱和图像, 包含丰富的空间信息与光谱信息, 可以实现无损批量地平面采集样本光谱信息。 该研究提出一种基于高光谱成像技术检测纸质文物狐斑的快速识别方法, 获取模拟纸质文物在360~970 nm的高光谱图像, 因360~450 nm受噪声影响过大, 所以选择剔除这部分光谱数据; 选取感兴趣区域并获取相应的平均光谱反射率, 比较健康区域与被狐斑感染区域, 发现两者的光谱曲线存在差异; 在450~600 nm附近, 受狐斑影响区域比健康区域的光谱反射率偏高, 并在600 nm附近出现波峰形态; 而在600~900 nm范围内, 被感染区域与健康区域的光谱都趋于平稳, 两者之间差异逐渐减小。 选取从特征波长对应的图像中提取的特征信息建立图像识别模型, 运用波段运算观察狐斑图像特征, 狐斑的大小和分布情况都能清晰地显示, 但与印章和墨迹重叠部分, 狐斑被印章和墨迹遮盖, 难以识别; 利用最小噪声分离, 虽然不同部分有重叠, 但能发现仅凭肉眼难以识别的隐藏的狐斑; 180条高光谱数据(450~970 nm)建立狐斑判别模型, 随机地分为120条数据为训练集, 60条数据为测试集, 应用K-近邻法与BP神经网络建立纸质文物狐斑光谱判别模型, 总体上两种方法对狐斑判别率分别达到73.3%和85%; BP神经网络相较于K-近邻模型, 总体判别率更高, 识别效果也更好。 结果表明, 利用高光谱成像可高效准确识别纸质文物狐斑, 为后续研究狐斑分布发展提供可靠的技术手段, 也为博物馆馆藏文物的保存提供指导意见。

Affected by preservation conditions, foxing will form on the surface of many paper cultural relics. If effective monitoring and scientific judgment are not carried out, the safety of paper cultural relics will be further affected. For the detection of foxing disease on paper cultural relics, there are problems such as hysteresis and subjectivity. It is difficult to identify the area covered by ink, paint and seals in the painting and calligraphy collection. Therefore, the concept of preventive protection based on cultural relics needs to be developed urgently. Non-destructive testing technology for efficient and accurate identification of foxing. The visible-near-infrared hyperspectral image combines spectrum and image, contains rich spatial information and spectral information, and can achieve lossless batch collection of sample spectral information on the flat. This research proposes a rapid identification method based on hyperspectral imaging technology to detect foxing on paper cultural relics. Obtain hyperspectral images of simulating paper cultural relics at the 360~970 nm. Because the 360~450 nm image is much affected by noise, we choose to exclude this part of the spectral data; select the region of interest, obtain the corresponding average spectral reflectivity, and compare the healthy area with that. In the area of foxing infection, it is found that there is a difference in the spectral curves of the two; near 450~600 nm, the spectral reflectivity of the affected area of foxing is higher than that of the healthy area, and the peak shape appears near 600 nm; and in the range of 600~900 nm, The spectrum of the infected area and the healthy area tends to be stable, and the difference between the two gradually decreases. Select the feature information extracted from the image corresponding to the feature wavelength to build an image recognition model, using band math to observe the image characteristics of foxing, the size and distribution of the foxing can be displayed, but the overlapping parts with the seal and ink, the foxing are covered by the seal and ink, which is difficult to identify; use the minimum noise fraction, although different parts are overlapping, it can find hidden foxing that is difficult to identify with the naked eye; 180 pieces of hyperspectral data (450~970 nm) establish a foxing discrimination model, randomly divided into 120 pieces of data as the training set, and 60 pieces of data as the test set, K-nearest neighbor method and BP neural network are used to establish a paper cultural relics foxing spectrum discrimination model. In general, the two methods have distinguished rates of 73.3% and 85% respectively; Comparing with the K-nearest neighbor model, the BP neural network has a higher overall discrimination rate and a better recognition effect. The results show that hyperspectral imaging can efficiently and accurately identify the foxing of paper cultural relics, provide reliable technical means for the follow-up research on the distribution and development of foxing, and provide guidance for the preservation of cultural relics in the museum.