One very interesting field of research in Pattern Recognition
that has gained much attention in recent times is Gesture
Recognition. In this paper, we consider a form of dynamic hand
gestures that are characterized by total movement of the hand (arm) in
space. For these types of gestures, the shape of the hand (palm) during
gesturing does not bear any significance. In our work, we propose
a model-based method for tracking hand motion in space, thereby
estimating the hand motion trajectory. We employ the dynamic time
warping (DTW) algorithm for time alignment and normalization of
spatio-temporal variations that exist among samples belonging to
the same gesture class. During training, one template trajectory and
one prototype feature vector are generated for every gesture class.
Features used in our work include some static and dynamic motion
trajectory features. Recognition is accomplished in two stages. In the
first stage, all unlikely gesture classes are eliminated by comparing
the input gesture trajectory to all the template trajectories. In the next
stage, feature vector extracted from the input gesture is compared
to all the class prototype feature vectors using a distance classifier.
Experimental results demonstrate that our proposed trajectory estimator
and classifier is suitable for Human Computer Interaction (HCI)
platform.['