The main contribution is a new approach called Random Hypersurface Model (RHM) for modeling the geometric shape of an extended target. An RHM assumes that each reflection center lies on a scaled version of the shape boundary, where the scaling factor is given by a random draw from a one-dimensional probability distribution. This model allows for a variety of target shapes reaching from basic shapes to arbitrary star-convex shapes by means of parameterizing the shape boundaries. Based on an RHM, it is possible to derive a Gaussian state estimator that recursively estimates the shape parameters and the kinematics of the The developed tracking method is evaluated in an example experiment, where the Microsoft Kinect sensor is employed as a Ground Moving Target Indicator (GMTI) target. The experiments show that the incorpomtion of the target extent significantly improves the overall tracking quality.
展开▼
