PRECISE TARGET TRACKING USING INFRA-RED MEASUREMENTS

In this paper the problem of tracking a target using a sequence of infrared (IR) images is addressed .The method is based on iteratively tracking a window center towards the target. The target image from the FLIR (forward looking infra-red) sensor is modeled as a bivariate Gaussian function, whose center position should be estimated .Based on this modeling for the radiation intensity pattern on the IR focal plane array, six different position estimation algorithms are proposed which are based on the centroid measurement, up sampling and curve fitting .After simulations the centroid measurement and upsampling which have the most accurate performance are chosen. A filter based tracker is also developed to exploit information of the motion characteristics of the target to improve the performance .Three type of filters are simulated; forth order kalman ,second order kalman filter and second order alfa-beta filter. In the forth order kalman filter, two different measurements are used, target position and its displacement between two successive frame. The simulation and experimental results show the validity of the performance predictions of the proposed algorithms

Keywords:

Infra-red Focal Plane Array (IR FPA) , centroid , least Squares fitting , up sampling , FLIR , filter , kalman filter.