基于光谱特征拟合的艺术画颜料成分识别研究  

Research on Art Painting Pigment Composition Recognition Based on Spectra Feature Fitting

作　　者：武锋强[1,2] 杨武年[1] 李丹[2,3] 

机构地区：[1]成都理工大学地质灾害防治与地质环境保护国家重点实验室,成都610059 [2]西南科技大学,绵阳621010 [3]中国工程物理研究院,绵阳621900

出　　处：《光散射学报》2014年第1期88-92,共5页The Journal of Light Scattering

基　　金：国家自然科学基金(41071265);四川省哲学社会科学重点研究基地民间文化研究中心项目(MJ12-05);西南科技大学青年基金(112X3109)

摘　　要：高光谱成像技术作为一种高效无损的检测方法,根据不同的物质成分在光谱上表现出不同的光谱信息,可以从图像上方便的鉴定出物质成分组成。本文分别利用可见光高光谱与红外高光谱成像技术对一幅古画进行颜料光谱成像,通过最小噪声分离(MNF)、纯净像元提取(PPI)、光谱特征拟合方法(SFF),与标准波谱库匹配,识别出该幅古画所用颜料成分主要由:朱砂、石青、白云母和金云母等组成,匹配度分别为:0.76,0.57,0.62,0.89。实验表明:利用高光谱成像技术不但可保存图像信息,物质成分波谱信息,还可还原受污染区域,为以后深层次研究提供详实资料。As an efficient lossless detecting method, hyperspectral imaging technology can easily identify the material composition from the image according to that the different material compo- nents represent of different spectral information. The spectrum of pigment on an ancient painting can be obtained by the visible light hyperspectral and infrared hyperspectral imaging technology, then it is found that the pigment mainly consists of cinnabar, azurite, muscovite and phlogopite by Minimum Noise Fraction Rotation （MNF）,Pixel Purity Index （PPI）, Spectra Feature Fitting （SFF） and matching of the standard spectrum library. The mathching index are 0.76,0.57,0.62 and 0.89. These experiments show that： this way is not only can save the image information, and the spec- trum information of material using hyperspectral imaging technology, but also can restore the origi- nal region, to provide the detailed data for advanced research.

关 键 词：高光谱 颜料鉴别 光谱特征匹配 

分 类 号：O433.4[机械工程—光学工程]