Ship recognition for improved persistent tracking with descriptor localization and compact representations

Sebastiaan P. van den Broek; Henri Bouma; Richard J. M. den Hollander; Henny E. T. Veerman; Koen W. Benoist; Piet B. W. Schwering

doi:10.1117/12.2067302

Translator Disclaimer

7 October 2014 Ship recognition for improved persistent tracking with descriptor localization and compact representations

Sebastiaan P. van den Broek, Henri Bouma, Richard J. M. den Hollander, Henny E. T. Veerman, Koen W. Benoist, Piet B. W. Schwering

Author Affiliations +

Proceedings Volume 9249, Electro-Optical and Infrared Systems: Technology and Applications XI; 92490N (2014) https://doi.org/10.1117/12.2067302
Event: SPIE Security + Defence, 2014, Amsterdam, Netherlands

Abstract

For maritime situational awareness, it is important to identify currently observed ships as earlier encounters. For example, past location and behavior analysis are useful to determine whether a ship is of interest in case of piracy and smuggling. It is beneficial to verify this with cameras at a distance, to avoid the costs of bringing an own asset closer to the ship. The focus of this paper is on ship recognition from electro-optical imagery. The main contribution is an analysis of the effect of using the combination of descriptor localization and compact representations. An evaluation is performed to assess the usefulness in persistent tracking, especially for larger intervals (i.e. re-identification of ships). From the evaluation on recordings of imagery, it is estimated how well the system discriminates between different ships.

Citation Download Citation

Sebastiaan P. van den Broek, Henri Bouma, Richard J. M. den Hollander, Henny E. T. Veerman, Koen W. Benoist, and Piet B. W. Schwering "Ship recognition for improved persistent tracking with descriptor localization and compact representations", Proc. SPIE 9249, Electro-Optical and Infrared Systems: Technology and Applications XI, 92490N (7 October 2014); https://doi.org/10.1117/12.2067302

ACCESS THE FULL ARTICLE