Abstract:
This paper presents a real-time multiple objects tracking system using multiple cameras. The objects used as targets are in spherical shape with almost identical diameter. In many applications, such object has been used as marker to be tracked to reduce complexity of object detection algorithm and help to simplify tracking process. The developed system has been implemented on a PC-cluster to improve real-time performance and provide system scalability. The cluster is constructed with two computers which data are transferred by using PCI-to-PCI data mover card. Processing modules have been developed to control system operations including image capturing, object detection, object identification and tracking, 3-D reconstruction, PC-to-PC communication, and camera calibration. After identifying corresponding objects in different views, 3-D coordinates of each target have been reconstructed in the main computer. Experimental results show that the system can track multiple objects moving in 3-D space in real-time. Tracking frequency can be increased with well prepared environment. Increasing light intensity in the measurement room or using active target, light inside spherical object, can be used to reduce camera exposure time. Furthermore, the system can be applied for specific tasks such as motion capture, virtual/augmented reality, or 3-D coordinate measuring for reverse engineering application when appropriate additional processing modules have been designed and added to the system.