Real-time image processing and analysis of highly dynamic objects

Аuthors

Kim N. V.^*, Bodunkov N. E.^**, Prokhorov P. D.^***

Moscow Aviation Institute (National Research University), 4, Volokolamskoe shosse, Moscow, А-80, GSP-3, 125993, Russia

*e-mail: nkim2011@list.ru
**e-mail: boduncov63@yandex.ru
***e-mail: prokhorov_pd@mail.ru

Abstract

The problem of surveillance and search of highly dynamic objects based on the image processing and analysis of video in real-time is considered in this paper. By highly dynamic objects, we mean such objects, which interframe shift exceeds their correlation radius. The realtime operation condition depends on the current task specifics, and can be expressed as a required dynamic objects computational efficiency (the amount of elementary operations) of the search algorithm.

As an example we examined search correlation algorithm MAD and search algorithm SIFT based on descriptors. We carried out studies of these algorithms and specify conditions under which these algorithms would fail to provide realtime condition fulfillment. The highly dynamic objects search algorithm ensuring realtime operation under these conditions is presenter in the paper. The advantage of the proposed algorithm lies in significant reduction of elementary operations due to required object descriptions presentation in the form of star pattern of characteristic points.

It is shown that proposed algorithm has higher computational efficiency than correlation algorithms and algorithms using descriptors.

We carried out experiments of real time objects tracking, and determination of the blade edge position of helicopter propeller. The experimental results show the efficiency and effectiveness of proposed algorithm application for tracking and tracing problems.

Keywords:

computer vision system, search algorithms and tracking handles, characteristic points.

References

1. Forsyth D. A., Ponce J. Computer vision a modern approach, Prentice Hall PTR, 2003, 928 p.
2. Vizilter Yu.V., Zheltov S.Yu., Bondarenko A.V., Osokov M.V., Morzhin A.V. Obrabotka i analiz izobrazhenii v zadachakh mashinnogo zreniya (Image processing and analysis for computer vision tasks), Moscow, Fizmatkniga, 2010, 671 p.
3. Abchuk V.A., Suzdal V.G. Poisk ob’ektov (Objects search), Moscow, Sovetskoe Radio, 1977, 334 p.