Visual location recognition using smartphone sensors for indoor environment

Vision based localization is an essential problem for many applications, such as augmented reality and pedestrian navigation. In this paper, a visual location recognition system based on multi-sensor information is proposed. In this system, a database called visual map is established before the localization stage. The visual map contains the visual features and the corresponding positional features. The basic idea of this localization method is to match two user´s query images with two database images by image retrieval and feature matching, and then we try to acquire the query camera poses by the use of the IMU and the electronic compass device on the smartphone. On the basis of the query camera poses and the positions of the matched features, a triangulation localization algorithm is introduced to determine the location of the user. The specific experiments for this visual location recognition system are performed, and the results demonstrate that this method can achieve the excellent performance in location recognition.