C. Video-based Tracking Techniques
With the availability of fast processors that capture video
frames efficiently, this type of technique becomes extremely
popular [24]. Being one of the most used [32] (when compared
with sensor-based and hybrid techniques) this technique can
determine the position of the camera relative to real-world
objects through image processing methods. Video-based techniques
are able to detect features in the environment. These
features can be artificially created by using markers (in which
a pattern is drawn) or can be natural patterns like human faces,
the latter being constantly used in ARZombie.
One possible approach consists of finding correspondence
between the characteristics of the captured image and its actual
coordinates. This operation can be done by using markers that
allow the graphics system to know the spatial coordinates and
orientation of the objects from the point of view of the user.
Some systems are built to analyze the contours and edges
in order to detect predefined models. The contours are the
most used features because they are robust to light changes.
One example of the usage of this method is to identify and
track, in real-time, known planar patterns, which consist of
a set of corners [19]. The main idea involves the continuous
monitoring of predefined patterns where, from the analysis of
each video frame, are detected not only the four outer corners
of the patterns as well as the interiors, in real time.
Other example [30] describes a new approach for detecting
and tracking multiple moving objects in video images acquired
by a static camera. Unlike many augmented reality applications,
this system, as well as the ARZombie game, does not
use extra tracking devices, only algorithms based on image and
video processing. The tracking system is able to track multiple
objects simultaneously. These PHOs (place holder objects) are
placed on a table and can be moved arbitrarily, allowing them
to be rotated independently of the viewing range of the camera,
which is oriented to the table.