Parameterization of the Acceleration Signals Recorded During Handwriting

This paper presents a novel method developed for preprocessing of the triaxial accelerometer output signals recorded during a handwriting. The DWT (Discrete Wavelet Transform) was applied to decompose acceleration of each axis into two components representing the acceleration caused by intended hand motion and the fluctuation of acceleration, respectively. Based on the first mentioned component various features were extracted to recognize handwriting and to characterize a person.