Abstract
The probability hypothesis density (PHD) filter, which is used for multi-target tracking based on sensor measurements, relies on the propagation of the first-order moment, or intensity function, of a point process. This algorithm assumes that targets behave independently, an hypothesis which may not hold in practice due to potential target interactions. In this paper, we construct a second-order PHD filter based on determinantal point processes which are able to model repulsion between targets. Such processes are characterized by their first- and second-order moments, which allows the algorithm to propagate variance and covariance information in addition to first-order target count estimates. Our approach relies on posterior moment formulas for the estimation of a general hidden point process after a thinning operation and a superposition with a Poisson point process, and on suitable approximation formulas in the determinantal point process setting. The repulsive properties of determinantal point processes apply to the modeling of negative correlation between distinct measurement domains. Monte Carlo simulations with correlation estimates are provided.
Similar content being viewed by others
References
Brezis H (1983) Analyse fonctionnelle. Collection Mathématiques Appliquées pour la Maîtrise. [Collection of Applied Mathematics for the Master’s Degree]. Masson, Paris
Clark D, de Melo F (2018) A linear-complexity second-order multi-object filter via factorial cumulants. In: 2018 21st international conference on information fusion (FUSION), pp 1250–1259
Clark D, Delande E, Houssineau J (2016) Basic concepts for multi-object estimation. Heriot-Watt University, Lecture notes
Clark D, Houssineau J (2012) Faa di Bruno’s formula for Gateaux differentials and interacting stochastic population processes. Preprint arXiv:1202.0264v4
Daley DJ, Vere-Jones D (2003) An introduction to the theory of point processes. Probability and its Applications, vol I. Springer, New York
de Melo F, Maskell S (2019) A CPHD approximation based on a discrete-gamma cardinality model. IEEE Trans Signal Process 67(2):336–350
Decreusefond L, Flint I, Privault N, Torrisi G (2016) Determinantal point processes. In: Peccati G, Reitzner M (eds) Stochastic analysis for Poisson point processes: Malliavin Calculus, Wiener–Itô chaos expansions and stochastic geometry, volume 7 of Bocconi & Springer series. Springer, Berlin, pp 311–342
Delande E, Üney M, Houssineau J, Clark D (2014) Regional variance for multi-object filtering. IEEE Trans Signal Process 62(13):3415–3428
Fränken D, Schmidt M, Ulmke M (2009) Spooky action at a distance in the cardinalized probability hypothesis density filter. IEEE Trans Aerosp Electron Syst 45(4):1657–1664
Georgii H, Yoo H (2005) Conditional intensity and Gibbsianness of determinantal point processes. J Stat Phys 118(1–2):55–84
Hoffman J, Mahler R (2004) Multitarget Bayes filtering via first-order multitarget moments. IEEE Trans Syst Man Cybern Part A Syst Hum 34(3):327–336
Hough J-B, Krishnapur M, Peres Y, Virág B (2009) Zeros of Gaussian analytic functions and determinantal point processes, volume 51 of University lecture series. American Mathematical Society, Providence
Jorquera F, Hernández S, Vergara D (2018) Multi target tracking using determinantal point processes. Progress in pattern recognition, image analysis, computer vision, and applications, volume 10657 of lecture notes in computer science. Springer, Berlin, pp 323–330
Jorquera F, Hernández S, Vergara D (2019) Probability hypothesis density filter using determinantal point processes for multi object tracking. Comput Vis Image Underst 183:33–41
Koch W (2018) On anti-symmetry in multiple target tracking. In: 2018 21st international conference on information fusion (FUSION), pp 957–964
Li T, Corchado J, Sun S, Fan H (2017) Multi-EAP: extended EAP for multi-estimate extraction for SMC-PHD filter. Chin J Aeronaut 30(1):368–379
Li T, Sattar TP, Han Q, Sun S (2013) Roughening methods to prevent sample impoverishment in the particle PHD filter. In: Proceedings of the 16th international conference on information fusion. IEEE, Istanbul, pp 17–22
Lund J, Rudemo M (2000) Models for point processes observed with noise. Biometrika 87(2):235–249
Macchi O (1975) The coincidence approach to stochastic point processes. Adv Appl Probab 7:83–122
Mahler R (2003) Multitarget Bayes filtering via first-order multitarget moments. IEEE Trans Aerosp Electron Syst 39(4):1152–1178
Mahler R (2007) PHD filters of higher order in target number. IEEE Trans Aerosp Electron Syst 43(4):1523–1543
Mahler R (2015) Tracking “bunching” multitarget correlations. In: IEEE international conference on multisensor fusion and integration for intelligent systems (MFI), pp 102–109
Mori S (1997) Random sets in data fusion. Multi-object state-estimation as a foundation of data fusion theory. In: Random sets (Minneapolis, MN, 1996), volume 97 of IMA volume in mathematics and its applications, pp 185–207. Springer, New York
Moyal J (1964) Multiplicative population processes. J Appl Probab 1:267–283
Moyal JE (1962) The general theory of stochastic population processes. Acta Math 108:1–31
Osada H (2013) Interacting Brownian motions in infinite dimensions with logarithmic interaction potentials. Ann Probab 41(1):1–49
Portenko N, Salehi H, Skorokhod A (1997) On optimal filtering of multitarget tracking systems based on point processes observations. Random Oper Stoch Equ 5(1):1–34
Schlangen I, Delande E, Houssineau J, Clark D (2018) A second-order PHD filter with mean and variance in target number. IEEE Trans Signal Process 66(1):48–63
Schumacher D, Vo B-T, Vo B-N (2008) A consistent metric for performance evaluation of multi-object filters. IEEE Trans Signal Process 56(8):3447–3457
Shirai T, Takahashi Y (2003) Random point fields associated with certain Fredholm determinants. I. Fermion, Poisson and boson point processes. J Funct Anal 205(2):414–463
Singh S, Vo B-N, Baddeley A, Zuyez S (2009) Filters for spatial point processes. SIAM J Control Optim 48(4):2275–2295
Soshnikov A (2000) Determinantal random point fields. Usp Mat Nauk 55(5(335)):107–160
van Lieshout MNM (1995) Stochastic geometry models in image analysis and spatial statistics, volume 108 of CWI Tract. Stichting Mathematisch Centrum, Centrum voor Wiskunde en Informatica, Amsterdam
Vo B-N, Ma W-K (2006) The Gaussian mixture probability hypothesis density filter. IEEE Trans Aerosp Electron Syst 54(11):4091–4104
Vo B-N, Singh SS, Doucet A (2005) Sequential Monte Carlo methods for multitarget filtering with random finite sets. IEEE Trans Aerosp Electron Syst 41(4):1224–1245
Vo B-T, Vo B-N, Cantoni A (2007) Analytic implementations of the cardinalized probability hypothesis density filter. IEEE Trans Signal Process 55(7):3553–3567
Vo B-T, Vo B-N, Cantoni A (2009) The cardinality balanced multi-target multi-Bernoulli filter and its implementations. IEEE Trans Signal Process 57(2):409–423
Acknowledgements
This research was supported by the Industrial Postgraduate Programme of the Economic Development Board of Singapore and Delphi Automotive Systems Singapore.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix: Janossy density approximation
Appendix: Janossy density approximation
Since the corrector terms \(l^{(1)}_{z_{1:m}} ( x )\), \(l^{(1)}_{z_{1:m}} (x;z)\), \(l^{(2)}_{z_{1:m}} (x,y)\), \(l^{(2)}_{z_{1:m}} (x,y;z)\), \(l^{(2)}_{z_{1:m}} (x,y;z,z')\) in (3.15), (3.19) and the kernel update formula (5.8) have no closed form expression in the determinantal setting, we propose to use the Janossy density approximations
\(n\ge 1\), which corresponds to a (Poisson) first-order approximation, and
\(n\ge 2\), which corresponds to a second-order (determinant) approximation, obtained from (4.8) by assuming that the off-diagonal entries \(J_\Phi (x_i,x_j)\), \(i\not = j\), are small.
This Janossy approximation is specially relevant to \(\alpha \)-determinantal Ginibre point processes (GPP) which approximate a Poisson point process when \(\alpha \in [-1,0)\) tends to 0, see Shirai and Takahashi [30].
Proposition A.1
Under (A.1) we have the first-order Poisson approximations \(l^{(1)}_{z_{1:m}} (x) \simeq J_\Phi (x,x)\), \(m\ge 0\), and
\(z\in z_{1:m}\), \(x\in \Lambda \), \(m\ge 1\).
Proof
by (3.9), which yields the approximation \(l^{(1)}_{z_{1:m}} ( x ) \simeq J_\Phi (x,x)\). On the other hand, for \(r=1,\ldots ,m\), using again (A.1) and (3.9) we have
We conclude by taking \(z_r=z\) and noting that by (3.15) and (A.3)–(A.4) we have
Proposition A.2
Under (A.1)–(A.2), we have the second-order approximations
\(z \in z_{1:m}\), \(x,y\in \Lambda \), \(m\ge 1\), and
\(z, z' \in z_{1:m}\), \(z\not = z'\), \(x,y\in \Lambda \), \(m\ge 2\), where
Proof
and for \(r,u=1,\ldots ,m\), using (A.1)–(A.2) and (3.9) we find
We conclude by taking \((z_r,z_u)=(z,z')\) and noting that by (3.20) and (A.6)–(A.7) we have
\(z, z' \in z_{1:m}\), \(z\not = z'\), \(m\ge 2\). Similarly, by (3.19) and (A.4), (A.6) we also have
As a consequence of (3.18) and Proposition A.2, the second-order conditional factorial moment density of \(\Phi \) given that \(\Xi =z_{1:m}=(z_1,\ldots , z_m)\) will be approximated as
\(m\ge 0\), with \(\rho ^{(2)}_{\Phi \mid \Xi =z_{1:m}} (x,x):=0\), \(x\in \Lambda \).
Proposition A.3
The (approximate) kernel update formula is given by
\(m\ge 0\), \(x,y\in \Lambda \).
Proof
By (3.14) and Proposition A.1, we have the approximation
hence by (A.8) and (A.9), we find
\(m\ge 0\), and we conclude by (4.4), i.e.
and (A.10).
The next result, which provides an approximation formula for the posterior covariance of Proposition 3.4, is a consequence of Proposition A.3 and (A.9).
Corollary A.4
Under (A.1)–(A.2) the posterior covariance of \(\Phi \) given that \(\Xi =z_{1:m}=(z_1,\ldots , z_m)\) is approximated as
1.1 Conclusion
Our observations have shown that the performance of the multi-target tracking PPP-based standard PHD filter is degraded in the presence of target interaction such as repulsion. To address this issue, we have constructed a second-order DPP-based PHD filter based on Determinantal Point Processes which are able to model repulsion between targets, and can propagate variance and covariance information in addition to first-order target count estimates. We have derived posterior moment formulas for the estimation of DPPs after thinning and superposition with a Poisson Point Process (PPP), based on suitable approximation formulas. Our numerical experiments include an assessment of the spooky effect on disjoint domains, with negative correlation estimates which are consistent with the nature of DPPs. We have also compared the robustness and performance recovery of the DPP and PPP-PHD filters when subjected to sudden changes in target numbers.
Rights and permissions
About this article
Cite this article
Privault, N., Teoh, T. Second-order multi-object filtering with target interaction using determinantal point processes. Math. Control Signals Syst. 32, 569–609 (2020). https://doi.org/10.1007/s00498-020-00271-x
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00498-020-00271-x
Keywords
- Probability hypothesis density (PHD) filter
- Higher-order statistics
- Correlation
- Second-order moment
- Determinantal point processes
- Multi-object filtering
- Multi-target tracking