TRACKING OBJECTS WITH UNKNOWN DYNAMICS Conference Paper uri icon

abstract

  • This paper discusses techniques that can be used by an imaging system to track an object that has unknown dynamics. A method to estimate an object's velocity from a single image using centroiding is presented. Inertial derivatives present one possible means of estimating motion by assuming linear dynamics. Accuracy of the estimate can be further improved by implementing a continuous-discrete Kalman filter on the measurements. A second method is then presented using non-rational Bézier curves to estimate the dynamics of the object in the image plane. Bézier curves are shown to be a superior method of tracking an object for short periods of time and a viable means of tracking an object exhibiting random walk behavior.

author list (cited authors)

  • Woodbury, D., & Mortari, D.

publication date

  • December 2009