The Understanding of Meaningful Events in Gesture-Based Interaction - Intelligent Video Event Analysis and Understanding - page 10

and operations on windows. The TAFFI (Thumb and Forefinger) interface avoids
complex hand tracking algorithms by simply detecting whether the two fingers are
touching and hence forming a closed ellipse-like shape. One and two-handed input
is demonstrated for various tasks.
Posture information can be successfully combined with location criteria. For ex-
ample, the Charade system of Baudel and Beaudouin-Lafon [2] allows a speaker
giving a presentation to control a remote computer with free-hand motions while
still using gestures in order to communicate with the audience. A numerical data
glove measures the amount by which fingers bend as well as the orientation of the
hand. Whenever the hand points in a given direction as specified by an active zone,
a cursor appears on the screen and starts following the hand motions. Gestures are
thus detected when the user's hand is pointing in the active area and actual seg-
mentation of the gesture command is performed using start and stop events defined
by specific wrist orientation and finger flexions. Charade is a good example of a
system that combines location with posture criteria in order to ease the segmenta-
tion process of gestures that can address different functions with different intentions
(e.g. controlling the screen and augmenting speech and discourse in front of the
audience).
Postures represent thus a powerful cue for segmenting motion gestures as they
resemble to some point the well-known and practiced paradigm of click-like inter-
action. The benefits from the system viewpoint are considerable: motion commands
are recorded and processed only when some specific postures are detected. This
avoids complex computations for analyzing movements trajectories especially at
various scales. The benefits are also important for the human operator as postures
create mental models of how the interface works.

2.3 Tap and Touch-Based Events

Touch-based interfaces have been emerging considerably in the last years. The mo-
tivation lies in the fact that tapping and touching represent intuitive operations that
allow direct contact with the object being selected or manipulated as well as hap-
tic feedback. It is interesting how tap and touch events are strongly connected to
interaction intent irrespectively of the technology being used for their detection.
Figure 7 illustrates this concept.
Surface computing and interactive tabletops have known a considerable devel-
opment with the technology used for capturing such touch events being resistive,
capacitive or surface wave [40]. For example, Dietz and Leigh [10] describe a tech-
nique for creating a touch-sensitive device that allows multiple users to interact si-
multaneously. Their surface generates modulated electric fields at each location that
are capacitively coupled to receivers present in the environment.
Computer vision has also been used for detecting touch and multi-touch events
by employing infrared cameras usually placed below a Plexiglas sheet illumi-
nated by an infrared light: when the finger touches the surface, the light becomes
frustrated meaning it scatters for the camera to detect it. The concept (frustrated