A Monocular Visual Positioning Method With Scale Information Using a Single Image Sensor

Monocular vision-based indoor positioning system has the advantages of simple structure, high-performance price ratio, and small calculation amount, which is an important technical support for visual indoor positioning. However, the scale ambiguity in monocular visual positioning causes a decrease in positioning accuracy, and using only absolute position estimation leads to more time consumption. To solve this problem, a monocular camera position estimation method with scale information is proposed to improve the positioning accuracy. First, the visual information is combined with inertial measurement unit (IMU) information to solve the scale factor of the translation vector and the initial position of the query camera. In addition, this article adopts the triangulation method and efficient perspective-n-point (EPnP) to estimate the relative position of the query camera to eliminate the time consumption caused by image retrieval in visual indoor positioning. Experimental results show that the proposed method can effectively improve the positioning accuracy.