Landmark Real-Time Recognition and Positioning for Pedestrian Navigation

摘要

The aim of this paper is to propose a new monocular-vision strategy for real-time positioning under augmented reality conditions. This is an important aspect to be solved in augmented reality (AR) based navigation in non-controlled environments. In this case, the position and orientation of the moving observer, who usually wears a head mounted display and a camera, must be calculated as accurately as possible in real time. The method is based on analyzing the properties of the projected image of a single pattern consisting of eight small dots which belong to a circle and one dot more at the center of it. Due to the simplicity of the pattern and the low computational cost in the image processing phase, the system is capable of working under on-line requirements. This paper presents a comparison of our strategy with other pose solutions which have been applied in AR or robotic environments.