Maneuvering Target Tracking with Image Enhanced Measurements

The problem of maneuvering target tracking is considered in this paper, with an additional (and non-conventional) target attitude measurement, i.e., fusion of data from a point detection sensor with an imaging sensor. This is motivated by the fact that certain attitude variables such as target orientation can provide faster and more reliable information about target maneuvers or acceleration changes than conventional position variables. An image based measurement channel of target orientation is characterized in discrete time and through it, the acceleration can be estimated. A modified extended Kalman filter that takes into account this image based acceleration estimation as well as its imperfection is developed. This filter has the property to alter the updating gain when maneuvers are observed. Simulation examples are presented to show the improved performance of position and velocity estimation.