Hyperspectral Technology Combined with Chemometrical Method for Distinction of Ink Type of Straight Liquid Ballpoint Pen

In order to fill the gap in the distinction of the ink type of the straight liquid ballpoint pen, a hyperspectral imaging test in the spectral interval of 400-1 000 nm was performed on the 29 common straight ballpoint ink samples from the market. The obtained spectral data was pretreated by simple smoothing, and a preliminary classification was carried out for the spectra according to their spectral trends. Further classification was made by using the Wald hierarchical clustering analytical method with SPSS.25 software, and its classification result was evaluated by using principal component analysis. It was found by results that the 29 ink samples were divided into the first category and the second category by the difference in the upward trend of the spectra at the spectra interval of 650-1 000 nm. Based on the dendrogram and scatter charts drawn according to the centralized plan of hierarchical clustering analysis, the 29 ink samples were further divided into 8 subcategories. 7 and 4 principal components of spectral data of the first and second categories were extracted by the principal component analysis, respectively, of which the cumulative variance contribution rate of the first two principal components reached 87% and 97% respectively. Score charts were drawn by values of the first two principal components, and it was shown that the samples in 8 subcategories could all be well clustered together, indicating that the effect of the hierarchinal clustering analysis was good.