The autonomous positioning of a vehicle predominantly relies on the global positioning system (GPS). However, in indoor environments, such as tunnels and indoor parking lots, the accuracy of GPS-based positioning can be significantly reduced due to weak GPS signals. To this end, we develop an accurate indoor vehicle positioning system using multiple fish-eye surveillance cameras. Our system first extracts vehicle segments from the top-view image of each fish-eye camera. These segments are then integrated into a common undistorted coordinate system. The center of the vehicle is finally determined using our simple but effective box fitting method. Moreover, a 1/18 scale indoor parking lot is designed to evaluate the performance of the proposed system. Throughout our experiments, we obtained average positioning errors of 30 or 24 cm in the regions covered by a single camera or multiple cameras, respectively.
Bibliographical noteFunding Information:
This work was supported in part by the Hyundai MNSOFT, Inc., in part by the Institute for Information and Communications Technology Promotion (IITP) grant funded by the Korea government (MSIT) under Grant 2017-0-00250
© 2007-2012 IEEE.
- Fish-eye camera
- image segmentation
- indoor navigation
- surveillance system
- vehicle positioning
ASJC Scopus subject areas
- Control and Systems Engineering
- Information Systems
- Computer Science Applications
- Computer Networks and Communications
- Electrical and Electronic Engineering