Online and Offline Data Collection of Japanese Handwriting
Yoko Seki · 2019
This paper describes online and offline Japanese character database developed for handwriting identification. A pair of online and offline data of Japanese handwritten characters were collected from four hundred and one participant using Anoto pen and Anoto pen paper. Participants were asked to write five sets of characters. Each set had seven hundred and two characters, which were the same all through the sets but were different in order. Online data was acquired by page, segmented into a character and used for the trajectory drawing of a handwriting. Calculation of writing speed and the difference in pen pressure between two measuring points were also acquired using the online data. Trajectory of each handwriting was acquired by connecting coordinates of measuring points and it was used for indicating stroke order, writing speed and the difference in pen pressure while writing along with indicating rough shape of the handwriting. Trajectories gave an image of the relationship between the writing behavior and the results of the behavior, that is, the rough shape of handwriting. Offline data was processed into image files by scanning at the resolution of 800 dpi by page and segmented into each handwritten character. This database enabled us to know visually intra and inter individual differences in handwritten characters and in the pattern of the pen pressure and the pen speed.