Kumar Raj, Kumar Vinay, Sharma Vishal
Institute of Forensic Science and Criminology, Panjab University, Chandigarh 160014, India.
Appl Spectrosc. 2015 Jun;69(6):714-20. doi: 10.1366/14-07663. Epub 2015 May 1.
Diffuse reflectance ultraviolet-visible-near-infrared (UV-Vis-NIR) spectroscopy is applied as a means of differentiating various types of writing, office, and photocopy papers (collected from stationery shops in India) on the basis of reflectance and absorbance spectra that otherwise seem to be almost alike in different illumination conditions. In order to minimize bias, spectra from both sides of paper were obtained. In addition, three spectra from three different locations (from one side) were recorded covering the upper, middle, and bottom portions of the paper sample, and the mean average reflectivity of both the sides was calculated. A significant difference was observed in mean average reflectivity of Side A and Side B of the paper using Student's pair >t-test. Three different approaches were used for discrimination: (1) qualitative features of the whole set of samples, (2) principal component analysis, and (3) a combination of both approaches. On the basis of the first approach, i.e., qualitative features, 96.49% discriminating power (DP) was observed, which shows highly significant results with the UV-Vis-NIR technique. In the second approach the discriminating power is further enhanced by incorporating the principal component analysis (PCA) statistical method, where this method describes each UV-Vis spectrum in a group through numerical loading values connected to the first few principal components. All components described 100% variance of the samples, but only the first three PCs are good enough to explain the variance (PC1 = 51.64%, PC2 = 47.52%, and PC3 = 0.54%) of the samples; i.e., the first three PCs described 99.70% of the data, whereas in the third approach, the four samples, C, G, K, and N, out of a total 19 samples, which were not differentiated using qualitative features (approach no. 1), were therefore subjected to PCA. The first two PCs described 99.37% of the spectral features. The discrimination was achieved by using a loading plot between PC1 and PC2. It is therefore concluded that maximum discrimination of writing, office, and photocopy paper could be achieved on the basis of the second approach. Hence, the present inexpensive analytical method can be appropriate for application to routine questioned document examination work in forensic laboratories because it provides nondestructive, quantitative, reliable, and repeatable results.
漫反射紫外-可见-近红外（UV-Vis-NIR）光谱法被用作一种手段，基于反射率和吸收光谱来区分各种类型的书写纸、办公用纸和复印纸（从印度文具店收集），而这些纸张在不同光照条件下的外观似乎几乎相同。为了尽量减少偏差，获取了纸张两面的光谱。此外，从纸张一侧的三个不同位置记录了三条光谱，覆盖纸张样本的上部、中部和底部，并计算了两面的平均反射率。使用学生配对t检验观察到纸张A面和B面的平均反射率存在显著差异。采用了三种不同的鉴别方法：（1）整套样本的定性特征，（2）主成分分析，以及（3）两种方法的结合。基于第一种方法，即定性特征，观察到鉴别能力（DP）为96.49%，这表明UV-Vis-NIR技术的结果非常显著。在第二种方法中，通过纳入主成分分析（PCA）统计方法进一步提高了鉴别能力，该方法通过与前几个主成分相关的数值载荷值来描述一组中的每个UV-Vis光谱。所有成分描述了样本的100%方差，但只有前三个主成分足以解释样本的方差（PC1 = 51.64%，PC2 = 47.52%，PC3 = 0.54%）；即前三个主成分描述了99.70%的数据，而在第三种方法中，总共19个样本中的四个样本C、G、K 和 N，在使用定性特征（方法1）时未被区分，因此对其进行了主成分分析。前两个主成分描述了99.37%的光谱特征。通过使用PC1和PC2之间的载荷图实现了鉴别。因此得出结论，基于第二种方法可以实现对书写纸、办公用纸和复印纸的最大鉴别。因此，这种廉价的分析方法适用于法医实验室的常规可疑文件检验工作，因为它提供了无损、定量、可靠且可重复的结果。
