A model-based approach to on-line cursive handwriting analysis and rec
ognition is presented and evaluated. In this model, on-line handwritin
g is considered as a modulation of a simple cycloidal pen motion, desc
ribed by two coupled oscillations with a constant linear drift along t
he line of the writing. By slow modulations of the amplitudes and phas
e lags of the two oscillators, a general pen trajectory can be efficie
ntly encoded. These parameters are then quantized into a small number
of values without altering the writing intelligibility. A general proc
edure for the estimation and quantization of these cycloidal motion pa
rameters for arbitrary handwriting is presented. The result is a discr
ete motor control representation of the continuous pen motion, via the
quantized levels of the model parameters. This motor control represen
tation enables successful word spotting and matching of cursive script
s. Our experiments clearly indicate the potential of this dynamic repr
esentation for complete cursive handwriting recognition.