New Augmented Reality Applications - Teaching through Multi-User Virtual Environments - page 368

that the virtual elements can be correctly rendered
and displayed.
Rendered elements are displayed on top of the
user's view of the real world. Mixing both this
view and the virtual graphics and text can be done
in either of two ways. See-through devices let the
user see the real world directly. They display the
graphics in a transparent screen located between
the real world and the user's eyes. The display
may be an LED or an OLED like those used in
projectors. Alternatively, a camera mounted on
the user's head may capture images of the real
world. These images may then be combined
with the virtual graphics and displayed on a head
mounted display (HMD). The main advantage of
AR systems over regular VEs is that they combine
virtual and real world imagery, thus providing a
much richer experience. Interaction between the
real and the virtual world is illustrated in Figure 1.
In order to develop Augmented Reality ap-
plications, there is a software library called AR-
Toolkit. The library is written in C/C++ and is
available under GNU's General Public License.
ARToolkit implements several computer vision
algorithms. These algorithms solve the main
problems of the technology, like computing the
camera location in real time, tracking a marker
or a real-world object, and estimating the camera's
intrinsic parameters. To solve these problems
ARToolkit uses pattern recognition techniques

applied to marker registration (see Figure 2). When
the system recognizes a marker, it retrieves infor-
mation about its size, orientation and position.
That information is then used to superimpose 3D
models on the marker's image.
Augmented Reality applications capture im-
ages from cameras. Once the system has an image,
it applies image processing algorithms to deter-
mine if the image contains a marker. Then, it
determines the marker's location in the image and
overlays one or more virtual objects on top of it.
Virtual objects are exactly aligned with the real-
world objects because the marker recognition
algorithm produces an accurate set of camera
parameters for the image. The final result makes
the users feel is if they were immersed in a hybrid
world.
AR systems can be single-user or multi-user
collaborative systems. Single-user systems have
been applied to science, engineering, training and
entertainment, among others. Collaborative sys-
tems have been applied to the same areas with much
more valuable results. For example, Schmalstieg
et al. (2002) developed the StudierStube system
to view scientific data. They use see-through
HMDs to allow multiple users to view the same
data superimposed on the real world.
Today AR is a matured technology. Early
surveys date back more than one decade (Azuma,
1997). Books have also been published entirely

Figure 1. Left, interaction between the real world and the virtual world in our AR system: a webcam
captures the real world, the computer combines it with the virtual objects, and the result is shown to
the user in the output screen; right, a screenshot of the real world overlaid with a chemical structure