Individual characteristics using pen writing behavior: intra- and inter-individual variability perspectives

Several personal authentication technologies are currently available. Writing movements are consistent among individuals, and each person has unique writing habits. Therefore, this study aims to evaluate the intra and inter-individual variability in pen angles to determine whether writing motions can be used for personal authentication. Sixteen right-handed adults participated in this study. Each participant was asked to write a Japanese name consisting of four kanji characters while seated on a chair. This task was repeated five times. Three-dimensional coordinate data were recorded from both ends of the pen using a motion-capture system. Four pen angles were calculated from the collected data: the horizontal plane angle, sagittal plane angle, frontal plane angle, and three-dimensional tilt angle. The angles were analyzed at the beginning of the first stroke of each character and at characteristic movements specific to Japanese kanji writing, such as "tome (stop)", "hane (upward brushstroke)", and "harai (sweeping stroke)". The standard deviation of the five trials was used as an index of intra-individual variability, while the standard deviation of the mean across the participants was used as an index of inter-individual variability. At all analyzed sites, the sagittal angle exhibited smaller intra-individual variability (1.16-1.63°) and larger inter-individual variability (7.46-9.18°) than the other angles. These results suggest that the sagittal plane may be effective for personal identification. At characteristic moments of movement, the horizontal plane angle was larger than the angle at the beginning of writing for both intra variability (2.90-4.29°) and inter-individual variability (11.10-18.92°). This trend was also observed for the other three pen angles. The conditions of small intra-individual variability and large inter-individual variability are ideal for individual identification. However, these findings suggest that the characteristic movements may not be suitable for personal authentication. Further investigation is required to identify the optimal writing motions for authentication purposes.