An approach to division into clusters of the objects on the IR-range images has been proposed. As the data for clustering a set of the key points-correspondences, obtained using the SIFT, SURF, ORB algorithms has been used. A typical object detection circuit has been compiled and the analysis of clustering algorithms has been performed. The constraints while using the existing approaches, which had pushed to creation of the clustering problem solution, have been revealed. The developed algorithm has been tested and has demonstrated its efficiency for the IR-range images. The results of the given approach to clustering can be used for classification of objects by area.
1. Rublee E., Rabaud V., Konolige K., Bradski G. ORB: an efficient alternative to SIFT or SURF. – California, 2012.
2. Jain A., Murty M., Flynn P. Data clustering: a review // ACM Computing Surveys. – 1999. – Vol. 31. – No. 3. – 264–323.
3. Котов А., Красильников Н. Кластеризация данных, 2006 [Электронный ресурс]. – Режим доступа: http://logic.pdmi.ras.ru/~yura/internet/02ia-seminar-note.pdf (дата обращения: 28.10.2016).
4. Николенко С. Кластеризация и алгоритм EM, 2012 [Электронный ресурс]. – Режим доступа: http://logic.pdmi.ras.ru/~sergey/teaching/mlau12/10-clustem.pdf (дата обращения: 28.10.2016).
5. Воронцов К.В. Алгоритмы кластеризации и многомерного шкалирования: курс лекций. – М.: МГУ, 2007. – 18 c.
6. Стрижов В.В., Кузнецов М.П., Рудаков К.В. Метрическая кластеризация последовательностей аминокислотных остатков в ранговых шкалах // Математическая биология и биоинформатика. – 2012. – Т. 7. – № 1.– С. 345–359.