Article
Semi-autonomous 3D tracking
Fecha
2012-06Autor
Andrade, Juan
Universidad de Cuenca
Dirección de Investigación de la Universidad de Cuenca
DIUC
Institución
Resumen
A 3D tracking system that works with a minimum of two cameras has been implemented. The proposed system consists of two main processes: a calibration process followed by a 3D tracking one. The calibration process is done only when the system is installed; but, should be repeated if camera parameters, either internal or external, are changed. Internal calibration was conducted based on the cameras’ final locations; therefore, internal parameters include operating conditions. The adopted Wide Baseline Matching (WBM) scheme provides feature descriptors with high distinctiveness. Matching is achieved by using a voting algorithm based on a similarity transform and the robust Random Sample Consensus (RANSAC) statistical method that enforces the epipolar constraints. The implemented WBM procedure provides feature correspondences between the image planes of the two cameras used for the external calibration. The 3D tracking process corresponds to the normal operation of the system after the calibration process. The proposed 3D tracking scheme which combines 2D tracking data from each camera is based on a triangulation method and the determined internal and external camera calibration parameters.