摘要
瞿昙寺因寺内所藏巨幅彩色壁画而闻名,但由于受到自然因素及人为因素的影响,寺内51间壁画廊中有些壁画已经发生褪色、龟裂、卷翘、起甲、颜料层脱落等病害,尤其是颜料层一旦脱落,对于不可再生的壁画来说,损失无法弥补。因此在壁画的保护修复过程中,对壁画病害的识别与评估就变得十分关键。以瞿昙寺西回廊的高光谱图像数据为研究数据,利用支持向量机(Support Vector Machine,SVM)方法对壁画病害识别分类;以此提取的病害矢量数据为基础,构建壁画病害的评估指标要素,进而提出基于多元统计回归的LS-SR-AIC评估模型,实现壁画病害严重程度的评估。实验结果表明:利用该方法颜料层脱落病害的SVM提取精度达95%以上,同时为颜料层脱落病害工程评价提供了一种新的方法指导,为文物的健康评价走向定量化评估提供参考与借鉴。
The temple of qutan is famous for its huge colored murals in the temple. However, due to natural and human factors, some murals in the 51 wall gallery in the temple have been fading, cracking, curling, lifting, and the paint loss. In particular, the disease of the paint loss occurs, and the loss cannot be compensated for the nonrenewable murals. Therefore, in the process of protection and restoration of murals, the identification and evaluation of mural diseases becomes critical. In this paper, the hyperspectral image data of the west corridor of the qutan Temple is used as the a test data. Spectral Angle Mapping (SAM) and Support Vector Machine (SVM) methods are used to identify the classification of mural diseases respectively. Based on the disease vector data, the evaluation index elements of mural diseases were constructed, and then the LS-SR-AIC evaluation model based on multivariate statistical regression was proposed to evaluate the severity of mural diseases. The experimental results show that the SVM extraction accuracy of the paint loss disease is more than 95%. At the same time, it provides a new way for the engineering evaluation of paint loss disease, and for the quantitative evaluation of the health evaluation of cultural relics.
作者
刘晓琴
侯妙乐
董友强
汪万福
吕书强
LIU Xiaoqin;HOU Miaole;DONG Youqiang;WANG Wanfu;LV Shuqiang(Beijing Advanced Innovation Center for Future Urban Design, Beijing University of Civil Engineering and Architecture, Beijing 100044, China;Beijing Key Laboratory for Architectural Heritage Fine Reconstruction & Health Monitoring, Beijing 100044, China;Beijing Soil & Water Conservation Ecology Engineering Consultants Co.,Ltd., Beijing 100055, China;Dunhuang Academy China, Gansu 730000, China)
出处
《地理信息世界》
2019年第5期22-28,共7页
Geomatics World
基金
北京市属高校高水平教师队伍建设支持计划长城学者培养计划项目(CIT&TCD20180322)
国家重点研发计划(NO.2017YFB1402105)资助
关键词
壁画病害
颜料层脱落
高光谱
指标要素
回归分析
mural disease
paint loss
hyperspectral imaging
index elements
regression analysis
