On Independent Axes Estimation for Extended Target Tracking

The trend towards high resolution sensos in combination with a growing number of automotive applications where precise estimates of dense near-range objects are required, results in an enormous need for high performance algorithms for tracking extended targets. Conventionally, this is soved by an ellipse shape approximation of the object extent. In this paper a novel method to estimate the shape parameters of an ellipse using multiple measurements is proposed. By means of an Eigenvalue Decomposition of the measurement spread matrix, the half axis can be measured. A Gaussian model for the feature observations is derived. The performance and consistency is shown by means of Monte Carlo simulations in comparison to state-of-the-art methods in literature.