کنفرانس مهندسی برق ایران

صفحه اصلی / سی و سومین کنفرانس بین المللی مهندسی برق

Image-Based Self-Localization Using Differential Observation Angle Based on Real-World Features

نویسندگان :

Seyed Mohammad Bagher Seyedin¹ Mahdi Goodarzi² Fereidoon Behnia³

1- Sharif university of technology 2- Sharif university of technology 3- Sharif university of technology

کلمات کلیدی :

Self-localization،camera pose estimation،differential observation angle،Quasi-Newton algorithm

چکیده :

In this paper, we propose a cost-effective and practical approach for two-dimensional image-based self-localization using a calibrated camera and identified Real-World Features (RWFs), such as natural landmarks and human-made structures. By mapping image pixels to observation angles through a straightforward calibration process, observation angles of RWFs presented in the image are obtained. Since the observation angles alone are insufficient without a known reference direction (e.g., north), the method leverages differential observation angles for localization. Using these differential angles and the known locations of the RWFs, we formulate an appropriate optimization problem, which is efficiently solved by a Quasi-Newton algorithm. We have evaluated the proposed approach in a practical scenario that involves three RWFs captured in a scene. The experimental results demonstrate sufficient performance and robustness against errors in calibration and pixel selection.