A novel validation method for high-precision IMU-based handwriting trace reconstruction

Abstract

The accurate reconstruction of handwriting gestures is crucial for both clinical and research applications. This study validates a Sensorized Ink Pen (SIP) reconstruction algorithm using an electromagnetic sensor as ground truth. Orientation, acceleration, and velocity signals were compared using RMSE metrics. Analyses were performed for “On-Sheet” and “In-Air” movements exhibiting larger discrepancies for rapid “In-Air” movements. Results show that the SIP reconstruction achieves mean positional errors of 3.5 mm. This validation framework ensures reliability for future handwriting-based assessments and machine-learning applications.