Algorithm for tracking objects on the sea surface by video surveillance with adaptive selection of sliding window size

by

B.A. Skorohod

 Sevastopol State University, RF, Sevastopol, Universitetskaya St., 33

E-mail: boris.skorohod@mail.ru

DOI: 10.33075/2220-5861-2023-1-118-126

UDC 004.9:004.41                                                                                            

Abstract:

   The article proposes a new adaptive algorithm for estimating the state and speed of marine objects on the sea surface by elevation and azimuth angles obtained from video image processing.  The algorithm is based on the combined use of an optimal filter with a sliding window and a detector for the appearance of uncontrolled disturbances, which allows you to choose the size of the sliding window based on the following considerations. There are several general requirements for its selection. Firstly, the model must be adequate to the object at intervals of sliding windows. Secondly, if the window size is too small, then the available information is not enough to obtain an acceptable accuracy estimate, and it should be increased. Conversely, if it is too large, then this may be unacceptable from the point of view of the transient characteristics of the filter. With this in mind, a large value is chosen in the absence of disturbances, and a small one during their action. The principal possibility to ensure the fulfillment of these conditions follows from the statement that the covariance matrix of the error in estimating an optimal filter with a finite impulse response is a monotonically non-decreasing matrix function. To illustrate the approach, we use chi-square statistics, widely used to solve various problems as a detector of such changes. This allows us to ensure theoretically the accuracy of estimates close to the Kalman filter (KF) in their absence. Numerical modeling has shown that the proposed algorithm has better transient characteristics compared to the KF and the optimal filter with a fixed sliding window size at the intervals of perturbation action and can provide close accuracy of estimates to the KF in their absence.

Keywords: filters with sliding window, azimuth and elevation angles, monocular video camera.

To quote: 

Full text in PDF(RUS)

REFERENCES

  1. Skorohod B.A. Receding Horizon Unbiased FIR Filters and Their Application to Sea Target Tracking, Journal of Control Science and Engineering, vol. 2018, Available: https://doi.org/10.1155/2018/1803623.
  2. Skorohod B.A., Stacenko A.V., and Fateev S.I. Algoritmy videonablyudeniya i manevrirovaniya avtonomnyh morskih sudov (Algorithms for video surveillance and maneuvering of autonomous marine vessels). Izvestiya TGU, tekhnicheskie nauki. 2018. No. 3, pp. 85–110.
  3. Bar-Shalom, Yaakov, Li, X Rong, Kirubarajan, Estimation with applications to tracking and navigation. Johh Wiley and Sons, New York, 2001.
  4. La Scala B.F., Mallick M., and Arulampalam S. Differential geometry measures of nonlinearity for filtering with nonlinear dynamic and linear measurement models. In SPIE Conference on Signal and DataProcessing of Small Targets, Vol. 6699, August 2007, 12 p.
  5. Shar P., and Li X.R. A practical approach to observability of bearings-only target tracking, SPIE Vol 3809, 1999, pp. 514–520.
  6. Ferdowsi M.H. Observability conditions for target states with bearing-only measurements in three-dimensional case. Proceedings of the 2006 IEEE International Conference on Control Applications Munich, Germany, October 4–6, 2006, pp. 1443–1449.
  7. Barbara La Scala Mark Morelande. An analysis of the single sensor bearings-only tracking problem 2008 11th International Conference on Information Fusion June 30, 2008, July 3, 2008, pp. 525–530.
  8. 8. Aidala V.J., Kalman filter behavior in bearings-only tracking applications. IEEE Trans. Aerosp. Electron. Syst., Jan. 1979, Vol. 15, pp. 29–39.
  9. X. Lin X., Kirubarajan T., Bar-Shalom Y., Maskell S. Comparison of EKF, pseudo measurement, and particle filters for a bearing-only target tracking problem, Proc. SPIE 4728, Signal and Data Processing of Small Targets 2002, pp. 240–250.
  10. Jazwinski A.H. Stochastic Processes and Filtering Theory. New York: Academic, 1970.
  11. Kwon W.H., Kim P.S., Han S.H. A receding horizon unbiased FIR filter for discrete-time state space models”, Automatica 38 (3) (2002), pp. 545–551.
  12. Shmaliy, Y.S., Liu, F., Zhao, S. & Khan, S. (2016). Unbiased, optimal, and in-betweens: the trade-off in discrete finite impulse response filtering. IET Signal Processing, 10(4), pp. 325–334. doi: 10.1049/iet-spr.2015.0360.
  13. Skorohod B. Diffuse Initialization of Kalman Filter, Journal of Automation and Information Sciences, V. 43. 2011, pp. 20–34. Begell House Publishing Ins., USA.
  14. Skorohod B. Diffuse Algorithms for Neural and Neuro-Fuzzy Networks: With Applications in Control Engineering and Signal Processing. United Kingdom1: Elsevier, 2017.
  15. Skorohod B. Covariance Analysis of the Receding Horizon Optimal FIR Filter. International Russian Automation Conference (RusAutoCon), September 2021. DOI: 10.1109/RusAutoCon52004.2021. 9537565.

Loading