中文摘要：

该文收集市售蓝色圆珠笔样品16份,结合模式识别技术对其高效液相色谱数据进行研究。利用相关系数法计算色谱图相对峰面积的相似度,定量地评价样品间的差异性。选取16个样品3个批次间的最小相似度（λ=0.92）作为判定是否源于同种样品的阈值,相似度小于0.92的数据为89份,占全部比对样品的74.2%;根据液相色谱图的保留时间、峰数及峰形差异将所有样品分为3类,进一步利用系统聚类法依据相似度的大小将第一类样品分为3小类。该文提出的采用相对峰面积相似度评价样品差异的方法,能够显著提高分辨能力,促进圆珠笔字迹的自动化识别。高效液相色谱法获取数据,模式识别技术分析数据的方法为圆珠笔字迹的分类与鉴别提供了新思路。

英文摘要：

Ink analysis is an important part of material evidence analysis in forensic science.It needs not only to acquire data by precision chemical analysis instruments,but also to parse data by the chemometrics methods such as principal component analysis（PCA）,cluster analysis（CA） and pattern recognition etc.16 kinds of blue ballpoint pen inks from the local market were studied by high performance liquid chromatography（HPLC） coupled with pattern recognition technology.With the similarities of the relative peak areas calculated by correlation coefficient,the differences of samples was evaluated quantitatively.We selected the minimum similarity（λ=0.92） from three batches of 16 samples as a threshold to assert if two samples came from the same kind of ballpoint pen ink.89 samples whose similarities were less than 0.92 accounted for 74.2% of all the combinations.All the ink samples were classified into three categories by the differences of retention time,peak number and peak shape.Then,the first category was divided into three subclasses according to disparities of their similarity degree by hierarchical cluster analysis（HCA）.Therefore,a more detailed classification was achieved.Results showed that the blue ballpoint pen inks which could not be identified by direct observation method were identified successfully and the similarities directly reflected the differences between samples,according to the similarity evaluation and HCA.The proposed method based on the similarity for the relative peak areas to evaluate samples could increase resolution significantly as compared with the direct observation method.It would provide a valuable tool for the classification and identification of ballpoint pen ink.