Omnidirectional Vision Tracking and Positioning for Vehicles

The target recognition, beacon tracking and object localization based on omni-directional vision (omni-vision) system are introduced in this paper. The development of omni-vision appears to have definite advantages for the performance of the mobile systems and real time control. We use a fisheye lens with the view angle of 185deg to build the omnidirectional vision system. A prototype of the unique landmark navigator for autonomous guided vehicle (AGV) based on the omni-vision system is presented. The special landmark integrates with spatial, color and shape features. Color features are extracted by the HSI color model. An omni-vision object tracking method based on improved mean shift algorithm is proposed. The positioning method just employs the view angle to the landmarks on geometric computation to estimate the position and orientation of the vehicle. The software flow chart based on the image process algorithm is provided. The hardware for an on-board navigator is fulfilled on the DM642 platform. The functions have been demonstrated in the laboratory experiments.