TY - GEN
T1 - Dynamic Handwriting Analysis of the Character ‘Y’ for Writer Profiling Using Geometric Principles and Ratios
AU - Slogrove, Kayleigh J.
AU - van der Haar, Dustin
N1 - Publisher Copyright:
© Springer Nature Singapore Pte Ltd 2020.
PY - 2020
Y1 - 2020
N2 - Graphology is the science of identifying a person or their emotional state based on their handwriting. Graphology is traditionally performed manually, the model proposed by this paper attempts to automate this process. The model proposed applies static handwriting algorithms to a dynamic handwriting system. The character written by the user is captured dynamically using a stylus. The static pipeline will focus on extracting geometric features of a written character and perform classification based on the ratios of the extracted features. This pipeline is then compared against the dynamic pipeline which uses the dynamic handwriting algorithm of dynamic time warping. The metrics obtained from testing the proposed system proves that the use of static algorithms on dynamic systems is still an accurate and valid classification method as the overall accuracy is significantly higher than the dynamic pipeline’s accuracy. The F1 scores for each class, within the static pipeline, further demonstrates that they were higher than the F1 scores from the dynamic pipeline. Furthermore, it was determined that the dynamic algorithm had the lower F1 score accuracy, however, this may be attributed to the small data sample used. It was found that the static features contributed more to the system than the dynamic features extracted.
AB - Graphology is the science of identifying a person or their emotional state based on their handwriting. Graphology is traditionally performed manually, the model proposed by this paper attempts to automate this process. The model proposed applies static handwriting algorithms to a dynamic handwriting system. The character written by the user is captured dynamically using a stylus. The static pipeline will focus on extracting geometric features of a written character and perform classification based on the ratios of the extracted features. This pipeline is then compared against the dynamic pipeline which uses the dynamic handwriting algorithm of dynamic time warping. The metrics obtained from testing the proposed system proves that the use of static algorithms on dynamic systems is still an accurate and valid classification method as the overall accuracy is significantly higher than the dynamic pipeline’s accuracy. The F1 scores for each class, within the static pipeline, further demonstrates that they were higher than the F1 scores from the dynamic pipeline. Furthermore, it was determined that the dynamic algorithm had the lower F1 score accuracy, however, this may be attributed to the small data sample used. It was found that the static features contributed more to the system than the dynamic features extracted.
KW - Dynamic
KW - Dynamic time warping
KW - Geometric principles
KW - Geometry
KW - Handwriting analysis
KW - Static
UR - http://www.scopus.com/inward/record.url?scp=85077492785&partnerID=8YFLogxK
U2 - 10.1007/978-981-15-1465-4_22
DO - 10.1007/978-981-15-1465-4_22
M3 - Conference contribution
AN - SCOPUS:85077492785
SN - 9789811514647
T3 - Lecture Notes in Electrical Engineering
SP - 211
EP - 222
BT - Information Science and Applications, ICISA 2019
A2 - Kim, Kuinam J.
A2 - Kim, Hye-Young
PB - Springer
T2 - 10th International Conference on Information Science and Applications, ICISA 2019
Y2 - 16 December 2019 through 18 December 2019
ER -