抄録
Development of high performance CPU, cameras and other sensors on mobile devices have been used for wide variety of applications. The applications require self-localization of the devices. Since the self-localization is based on GPS, gyro sensor, acceleration meter and magnetic field sensor (POS) of low accuracy, the applications are limited. On the other hand, self-localization method using images have been developed, and the accuracy of the method is increasing. This paper develops the self-localization method by integrating sensors, such as POS and a camera on a mobile device. The proposed method mainly consists of two parts : one is the accuracy improvement of POS data with filtering, and another is development of self-localization method by integrating POS data and images. The POS data filtering combines all data by using Kalman filter. The estimated exterior orientation factors are used as initial value of ones in image-based self-localization method, which is based on structure from motion. The exterior orientation factors are integrated and updated by applying bundle adjustment. Through experiments with real data, the accuracy improvement by the proposed method is confirmed.
