Comparison of signatures on paper and graphic pad using multivariate analysis

• Main Author: Alagan, Vinothiniy
• Format: Thesis
• Language: English
• Published: 2022
• Subjects: HV Social pathology. Social and public welfare
• Online Access: http://eprints.usm.my/58533/1/VINOTHINIY%20AP%20ALAGAN-24%20pages.pdf; http://eprints.usm.my/58533/

Most sectors are progressively adopting electronic signatures as a standard business method due to technology improvements and the allure of a paperless workflow. The increasing use of electronic signatures has presented challenges for document examiners in their methods of examination due to the absence of well defined procedures, pertinent research, and differences between electronic signatures and handwritten signatures. As a first step to answer the question regarding the differences in signature writers in various mediums, 30 signatures were collected on Double-A, A4 paper using a blue Ballpoint pen Grip X10 1.0mm, and another 30 signatures on the XP-Pen Star G430 graphic pen tablet with a P01 Stylus pen from 5 subjects. Then, the similarities and differences between the signatures were analysed using three parameters that represent vertical dimension movement in writing, namely the ratio of the upper zone to the total height, the ratio of the middle zone to the total height and the ratio of the lower zone to the total height. The parameters were then tested with multiple tests. The ratios were used to calculate the mean and RSD and subsequently for statistical analysis. In summary, intra-variation among the authors was evident, regardless of the writing medium used in signatures. These suggested that natural variation was common, and no people could produce the same signature every occasion. Based on the statistical results, the K-mean clustering accuracy was at least 60% and above for all the signees. Perhaps, this shows differences in the signature written on paper and graphic pad, although the same person wrote it simultaneously. In factor analysis followed by scatter plots, the signatures analysed in this research are exhibited within and between varying writing mediums. The within variation suggested that natural variation was expected regardless of the writing medium used in signatures, and no people could produce the same signature on every occasion. The variation between writing mediums, namely paper and graphic pad, is evident, although the same person signs it. Therefore, the null hypothesis is rejected, and the alternative hypothesis is accepted where the signature from the same person will be significantly different when using different writing mediums.