Camera-Based Mobile Indoor Guidance System in Mixed Reality

We present a mobile augmented reality (AR) system for indoor guidance where we apply vision technique without using markers. There are two main problems: First, to augment suitable information according to user's situation, it has to be identified which place the user belongs to. Second, to put information into the scene and to make it aligned with the target scene element, the basic structure of the augmenting space should be grasped. For place recognition in real time mobile system, we employ simple feature detection method combined with graph based spatial connectivity representation. Also, image-based analysis method is applied to interpret the basic scene structure from the video.