Search

A simple yet effective state-estimation algorithm is presented and demonstrated to have advantages over previous standard clustering techniques used for the particle probability hypothesis density filter. The idea behind the proposed algorithm is that it uses the latest available information (i.e., the measurements) to direct particle clustering. The particle likelihood and target number estimation, computed during probability hypothesis density recursion, are both used to partition particles into clusters, and the center of each cluster gives the state estimation of an individual target. Simulation results indicate that the proposed algorithm outperforms the standard clustering approach using the k-means algorithm, achieving higher accuracy and shorter computational time.

For solving the problem in space object passive tracking with too short arcs acquired in space-based optical surveillance mission, the concepts of the attributable and the admissible region are introduced. And a triangulation subdivision and iterative searching method is put forward based on the triangulation node sorting by the cost function values. Simulation results indicate that this new method can converge rapidly and effectively, and the ill-condition in the space object tracking with too short arcs can also be avoided.

The problem of super-resolution and tracking for midcourse closely spaced objects (CSO) is examined using a space-based infrared sensor. Within a short time window, the midcourse CSO trajectories on the focal plane can be modeled as following a straight line with a constant velocity. Thus, the object's initial state (location and velocity on the focal plane) exclusively corresponds to its trajectory on the focal plane. Thereupon, the objects number, intensities and initial states, as well as the sensor noise variances, are considered random variables, and a Bayesian model is proposed which is utilized to define a posterior distribution on the joint parameter space. To maximize this distribution, reversible jump Markov chain Monte Carlo algorithm is adopted to perform the Bayesian computation. The proposed approach simultaneously used the multiple time-consecutive frame data to estimate model parameters. Compared with the single-frame method, it not only gains the super-resolution capability but also can directly estimate focal plane trajectories without using explicit data association techniques. Results show that the performance (estimation precision of objects number, focal plane locations, intensities and ballistic trajectories for the CSO, together with the computation load) of the proposed approach outperforms the conventional single-frame and multiframe approaches.

This paper presents a novel approach to tracking a large number of closely spaced objects (CSO) in image sequences that is based on the particle probability hypothesis density (PHD) filter and multiassignment data association. First, the particle PHD filter is adopted to eliminate most of the clutters and to estimate multitarget states. In the particle PHD filter, a noniterative multitarget estimation technique is introduced to reliably estimate multitarget states, and an improved birth particle sampling scheme is present to effectively acquire targets among clutters. Then, an integrated track management method is proposed to realize multitarget track continuity. The core of the track management is the track-to-estimation multiassignment association, which relaxes the traditional one-to-one data association restriction due to the unresolved focal plane CSO measurements. Meanwhile, a unified technique of multiple consecutive misses for track deletion is used jointly to cope with the sensitivity of the PHD filter to the missed detections and to eliminate false alarms further, as well as to initiate tracks of large numbers of CSO. Finally, results of two simulations and one experiment show that the proposed approach is feasible and efficient.