2. RESEARCH METHOD
There are many methods available for marker detection and tracking in Augmented Reality,
Taketomi et al [11] describe a robust outdoor tracking method that uses two stages. In the initial off line
stage, a landmark database is constructed from structure from motion data and laser range finder information.
After this database is constructed, landmark tracking can be used in a non-line stage to provide wide-area
outdoor tracking for mobile AR applications. Stephen DiVerdi uses a vision-based approach for orientation
tracking [12]. It tracks the camera orientation in real time and simultaneously creates an environment map by
calculating the optical flow between successive frames. These measurements are refined with more
computationally expensive landmark tracking to avoid the drift that frame-to-frame feature matching
introduces. But this approach can’t run on phones due to high computational costs, because the method
requires extensive GPU. Sensor based tracking obtain geo-location information by the fusion of several
different sensors like GPS, magnetometer and linear accelerometers to determine where annotations should
appear in the camera view. AR also needs an internet connection to receive data for the information layer.