ADAPTIVE MATRIX MODELS IN THE VIDEO STREAMS CONTROL PROBLEM

Authors

  • S.V. Mashtalir Kharkiv National University of Radio Electronics, Kharkiv, Ukraine
  • M.I. Stolbovyi Kharkiv National University of Radio Electronics, Kharkiv, Ukraine

DOI:

https://doi.org/10.15588/1607-3274-2018-4-18

Keywords:

video data, clustering, adaptive matrix models, tuning criteria, tracking signal.

Abstract

Context. At present, the multidimensional data analysis is one of the priority scientific research areas. This is due to the almost
uncontrollable growth in the information size and the need to obtain/search for various kinds of useful data from it. At the same time,
video data analysis is one of the most difficult from a computational point of view, not only because of BigData being processed, but
also due to the video unstructuredness, and also the fact that in a bunch of video processing applications exist limitations on the
processing time. One of the ways to solve these video analysis problems is to pre-process the initial data in order to get them split
into homogeneous segments (shots), which significantly reduces the time and computational costs for further content-based video
analysis in video database. And, despite the existing results in this direction, the video sequences clustering/segmentation problem
remains extremely relevant.
Objective. The paper considers the problem of clustering multidimensional streaming data as example of temporal video
segmentation.
Method. A method for controlling changes in streaming data is proposed, which allows you to detect the moments of a
significant change in the input multidimensional data characteristics, based on adaptive matrix models with the specialized tuning
algorithm for the predictive model introduction.
Results. The conducted experiment on an arbitrary video sequences demonstrated the video shot detection possibility. It should
be noted that the proposed approach essentially depends on the input data spatial segmentation results, which is necessary to obtain a
characteristics set describing each frame of the video sequence.
Conclusions. The proposed method allows multidimensional input data clustering/segmentation by adaptive matrix models. As
initial data in the experimental part, video sequences were used.

References

Grundmann M., Kwatra V., Han M., et al. Efficient hierarchical graph-based video segmentation, IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2010, pp. 2141–2148. DOI: 10.1109/CVPR.2010.5539893

Galasso F., Cipolla R., Schiele B. Video Segmentation with Superpixels, 11-th Asian Conference on Computer Vision (ACCV), 2012, Volume I, pp. 760–774. DOI:

https://doi.org/10.1007/978-3-642-37331-2_57

Seguin G., Bojanowski P., Lajugie R. et al Instance-level video segmentation from object tracks, IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016, pp. 3678–3687. DOI: 10.1109/CVPR. 2016.400 4. Abonyi J., Feil B., Nemeth S. et al. Fuzzy clustering based segmentation of timeseries, Lecture Notes in Computer Science. Berlin, Springer, 2003, Vol. 2810, pp. 275–285. DOI:https://doi.org/10.1007/978-3-540-45231-7_26

Abonyi J., Feil B., Nemett S. et al. Modified Gath-Geva

clustering for fuzzy segmentation of multivariate timeseries, Fuzzy Sets and Systems, 2005, Vol. 149, Issue 1, pp. 39–56. DOI: 10.1016/j.fss.2004.07.008

Badavas P. C. Real-time statistical process control. Englewood Cliffs, N. J., Prentice Hall, 1993, 232 p.

Bodjans’kyj Je. V., Peleshko D. D., Vynokurova O. A. et al Analiz ta obrobka danyh zasobamy obchisluvalnogo intelektu. L’viv, Vyd-vo L’vivs’koi’ politehniky, 2016, 236 p. ISBN: 978-617-607-902-6

Hoeppner F., Klawonn, F. Fuzzy clustering of sampled functions, In: Proc. 19th Int. Conf. of the North American Fuzzy Information Processing Society (NAFIPS). Atlanta, USA. 2000, pp. 251–255. DOI: 10.1109/NAFIPS. 2000.877431

Mashtalir S., Mashtalir V., Stolbovyi M. Video shot

boundary detection via sequential clustering, International Journal “Information Theories and Applications”, 2017, Vol. 24, Number 1, pp. 50–59.

Mantula E. V., Mashtalir S. V. Matrichnaja

prognozirujushhaja model’ i ee obuchenie v zadache

jekologicheskogo monitoringa, Jelektrotehnicheskie i

komp’juternye sistemy, 2013, No. 10(86), pp. 152–156.

Bodjanskij E. V., Pliss I. P. O reshenii zadachi upravlenija matrichnym ob’ektom v uslovijah neopredelennosti, Avtomatika i telemehanika, 1990, No. 2, pp. 175–178.

Chuev Ju. V., Mihajlov Ju. B., Kuz’min V. I. Prognozirovanie kolichestvennyh harakteristik processov. Moscow, Sov. Radio, 1975, 400 p.

Trigg D. W., Leach A. G. Exponential smoothing with an adaptive response rate, Operational Research Quarterly, 1967, 18, No. 1, pp. 53–59.

Mashtalir S. V. Mnogomernoe jeksponencial’noe sglazhivanie v zadachah analiza videodannyh, VI Mіzhnarodna shkola-semіnar «Teorіja prijnjattja rіshen’». Pracі shkolisemіnaru. Uzhgorod, UzhNU, 2012, pp. 136–137.

How to Cite

Mashtalir, S., & Stolbovyi, M. (2019). ADAPTIVE MATRIX MODELS IN THE VIDEO STREAMS CONTROL PROBLEM. Radio Electronics, Computer Science, Control, (4). https://doi.org/10.15588/1607-3274-2018-4-18

Issue

Section

Progressive information technologies