For tracking a target in a heavily cluttered environment, the Probabilistic Data Association Filter (PDAF) is very efficient and can significantly reduce track losses. However, as shown in this paper, the PDAF will experience difficulties at the initial stage of the filtering when the track is not accuracy enough, and the filter tends to diverge under even modest clutter density. To address this problem we propose a technique of splitting the track of the target into sub-tracks that run in parallel when the original track has low accuracy. Each sub-track occupies a portion of the uncertainty region of the original track. As a result, the sub-tracks maintained using the PDAF will be more selective over the incoming measurements (including detection and false alarms), and have less loss in tracking accuracy and improved robustness. This approach is similar to the Gaussian Sum filter in the literature. The major contribution of this paper is to propose a systematic method to effectively divide a less accurate track in a high dimensional state space into a set of sub-tracks to effectively improve the robustness of the PDAF. The splitting of the track will incur a significant amount of additional computation cost. To reduce the number of sub-tracks a likelihood ratio test is also proposed for the problem considered to drop unlikely sub-tracks. Simulation results are presented to demonstrate the performance of the proposed algorithm.
[1]
Peter Willett,et al.
GMTI Tracking via the Gaussian Mixture Cardinalized Probability Hypothesis Density Filter
,
2010,
IEEE Transactions on Aerospace and Electronic Systems.
[2]
Fred Daum,et al.
Particle flow for nonlinear filters with log-homotopy
,
2008,
SPIE Defense + Commercial Sensing.
[3]
Friedrich Faubel,et al.
The Split and Merge Unscented Gaussian Mixture Filter
,
2009,
IEEE Signal Processing Letters.
[4]
Y. Bar-Shalom,et al.
Tracking in a cluttered environment with probabilistic data association
,
1975,
Autom..
[5]
H. M. Shertukde,et al.
Performance evaluation of a cascaded logic for track formation in clutter
,
1989
.
[6]
J. Bather,et al.
Tracking and data fusion
,
2001
.
[7]
Thiagalingam Kirubarajan,et al.
Estimation with Applications to Tracking and Navigation
,
2001
.
[8]
H. Sorenson,et al.
Recursive bayesian estimation using gaussian sums
,
1971
.
[9]
Petar M. Djuric,et al.
Gaussian sum particle filtering
,
2003,
IEEE Trans. Signal Process..
[10]
H. Sorenson,et al.
Nonlinear Bayesian estimation using Gaussian sum approximations
,
1972
.