Passive Tracking for Mobile Augmented Reality Using Landmark Identification

Abstract

Augmented Reality is to integrate artificial elements such as sound, graphics, images, video and tactile feedback into a user s real world environment in order to improve the user s perception of the reality. Nowadays sophisticated mobile devices that consist of 3D rendering capabilities and various kinds of sensors built into them have become part of our lives. With this growing complexity and performance of mobile devices and their wide usage and popularity, the augmented reality has shifted towards mobile computing. When considering tracking and registration in a mobile augmented reality system, as virtual objects are placed in the field of view of the user; the tracking should be done in real time with low latency. The jitter should also be low enough so that user may not see augmented objects changing their position with user s movements. Most vital is the techniques used need to be computationally less complex to process in mobile platforms. In order to augment computer generated graphics objects seamlessly into the real world images, an accurate computation of the user s pose with respect to the virtual world is required. The existing tracking methods such as GPS, small sensor based methods, and image based and vision based methods currently in existence has some limitations. High cost of sensor based methods, requirement of artificial feducials in image based methods are some of these limitations. In order to counter some of these limitations, we propose a landmark based tracking technique. The proposed method does not use any artificial markers. It uses set of images of the planer surfaces in the environment to determine the pose. Experimental results show that this tracking method is accurate, has low jitter and can perform at 10 Hz speed.