Human and Virtual Agent Expressive Gesture Quality Analysis and Synthesis - Coverbal Synchrony in Human-Machine Interaction - page 287

Table 1. A comparison of different approaches for expressive gesture quality analysis and

synthesis.

Measure Analysis Synthesis

Spatial
aspects of
movement

maximum distance of hand and elbow
(Bernhardt and Robinson, 2007)
distance between two hands (Cardakis et
al., 2007)
minimum rectangle surrounding the body
(contraction index, Mancini and Castellano,
2007)

joint distance from the body
(Hartmann et al., 2005), (Neff
and Fiume, 2003)

Temporal
aspects of
movement

average hand and elbow speed (Bernhardt
and Robinson, 2007)
hand barycenter velocity (Mancini and
Castellano, 2007)

gesture phase duration
computed according to the
Fitt's law (Hartmann et al.,
2005)

Fluidity,
Smoothness

sum of variance of the norms of the motion
vectors (Caridakis et al., 2007)
approximated by the Directness: ratio
between the length of the shortest line and
the actual trajectory (Camurri et al., 2004a)
discrepancy between movements of
different body parts (Mazzarino et al., 2007)
relation of the sequence of motion and
non-motion phases (Mazzarino et al., 2007)

sequential joint activation (Neff
and Fiume, 2003)
duration of and amplitude of
the preparation and overshoot
phase (Szczuko et al., 2009)
trajectory interpolation
procedure (Hartmann et al.,
2005)
concatenation of two gestures
(Hartmann et al., 2005)

Energy,
Power,
Impulsivity

hand and elbow acceleration (Bernhardt
and Robinson, 2007)
fi rst derivate of motion vectors (Caridakis
et al., 2007)
hand acceleration (Mancini and Castellano,
2007)
peak in the time series of quantity of motion
(Mazzarino et al., 2007)
kinetic energy of user's joints (Piana et
al., 2012)

tension and bias of TCB spline
lines (Harmann et al., 2005)

Tension,
Stiffness

timing and amplitude of
gesture phases (Ness and
Fiume, 2002)

Overall
activity

sum of the motion vectors (Cardakis et
al., 2007)
Quantity of motion, velocity (Mancini and
Castellano, 2007)

Threshold-based gesture
selection algorithm (Hartmann
et al., 2005)

and Jennifer Hofmann from University of Zurich (Switzerland) for
their valuable comments and suggestions.
This research has been partially funded by the European Community
Seventh Framework Programme (FP7/2007-2013) ICT, under grant