Preliminary Validation of a Cursive Handwriting Reconstruction Algorithm from a Sensorized Ink Pen

Authors: Toffoli, S., Gentile, C., Dui, L. G., Fontolan, S., Termine, C. Ferrante, S

Keywords

Handwriting reconstruction; Sensorized ink pen; IMU; Optical Character Recognition; Signal processing; Digital health; Dysgraphia assessment; Kinematic analysis; Sensor Fusion.

Summary

This article proposes and preliminarily validates an algorithm for reconstructing cursive handwriting based on data collected by a sensor-equipped pen (Smart Ink Pen), which records movement (IMU) and force signals whilst writing on paper.
The aim is not only to obtain information about the writing process, but also to reconstruct the final graphic trace, whilst preserving the naturalness of the act (unlike tablets). The algorithm reconstructs the trajectory of the pen tip through signal integration and spatial transformations.
The quality of the reconstruction was assessed using an optical character recognition (OCR) system, by comparing the reconstructions with actual scans. The results show lower performance compared to the scans (CRR ~59% vs ~81%), but are nonetheless promising.
Despite this, the work demonstrates the feasibility of reconstructing handwriting from sensory signals, paving the way for quantitative tools for the assessment of handwriting (e.g. dysgraphia) in real-world contexts.

For further details, here is the link for the open access publication.