Abstract
In this paper, we propose a novel indoor human navigation system by applying view-based navigation to a smartphone. Indoor human navigation systems should conventionally need to install positioning infrastructures in buildings. To solve this problem, we focus on the view-based navigation developed as an indoor positioning method for mobile robots. This method has the following two features. One is able to identify the position of a user by processing images obtained from a monocular camera. The other is profitable in the processing cost because the process for position identification can be realized by a simple image matching algorithm. By using these two features, we will show that a human navigation can be implemented on a smartphone without installing special infrastructures. For applying the view-based navigation to a smartphone, we optimize an image matching algorithm and we also visualize the positional information obtained by the image matching. In the experiments of human navigation in actual indoor environments like a corridor, we confirmed that the proposed system is capable of accomplishing the indoor human navigation on a smart phone.
