Reconstruction of charged tracks in the presence of large amounts of background and noise

We consider the problem of track reconstruction with very large amounts of background and noise. This paper is the sequel to a previous publication [R. Frühwirth, A. Strandlie, Nucl. Instr. and Meth. A, to appear], which deals with track finding in various noise scenarios. We extend the study by considering the full track reconstruction chain, i.e., by submitting the track candidates from the track finding procedure to a track fit for obtaining final estimates of the track parameters. It is shown that a standard least-squares estimator such as the Kalman filter does not reach optimal precision after the track fit and that the estimates can be improved by using the Deterministic Annealing Filter (DAF) for the final hit assignment and parameter estimation.