Ant colony optimization algorithm for object identification in multi-cameras video tracking systems / Krzysztof Schiff // W: New advances in dependability of networks and systems : proceedings of the Seventeenth ... więcej

Ant colony optimization algorithm for object identification in multi-cameras video tracking systems / Krzysztof Schiff // W: New advances in dependability of networks and systems : proceedings of the Seventeenth International Conference on Dependability of Computer Systems : DepCoS-RELCOMEX, June 27-July 1, 2022, Wrocław, Poland [Dokument elektroniczny] / eds. Wojciech Zamojski [et al.]. – Cham : Springer, 2022. – (Lecture Notes in Networks and Systems, ISSN 2367-3389 ; 484). – S. 276-286 : rys., tab. – Tryb dostępu: https://link.springer.com/chapter/10.1007/978-3-031-06746-4_27. – Bibliogr. 44 poz., Abstr. – doi: 10.1007/978-3-031-06746-4_27. – ISBN 978-3-031-06746-4 (online). – ISBN 978-3-031-06745-7 (print)

The multi-dimensional assignment is one of methods used to associate measurement data with tracks in the analysis of image frames in vision systems. This is a very difficult problem and belongs to the ... więcej

The multi-dimensional assignment is one of methods used to associate measurement data with tracks in the analysis of image frames in vision systems. This is a very difficult problem and belongs to the hard NP class. Since accurate methods are very time consuming, there is a need for develop other methods to obtain solutions to this problem in a reasonable time. Among them there is the ant colony optimization heuristic (ACO). ACO has already been proposed for this problem, but only with a static function of desire. The article presents the dynamic function of desire together with data structures and discusses the results of the conducted experiments. The ant algorithm with the dynamic function of desire shows its superiority over the ant algorithm with the static function of desire because it gives a better assignment each time the algorithm is run. The proposed ant algorithm with the new function of desire is not time consuming and can be implemented in multi-cameras vision systems for objects tracking.