10 likes | 322 Views
Domenico Daniele Bloisi and Luca Iocchi Sapienza University of Rome Dipartimento di Informatica e Sistemistica via Ariosto 25, 00185 Rome, Italy. Rek-means A k-means Based Clustering Algorithm.
E N D
Domenico Daniele Bloisi and Luca IocchiSapienza University of RomeDipartimento di Informatica e Sistemisticavia Ariosto 25, 00185 Rome, Italy Rek-means A k-means Based Clustering Algorithm • The Algorithm: Rek-means is a new clustering method based on k-means that has been implemented on the ARGOS video surveillance system in the city of Venice. • K-means limitations: in using k-means, one encounters a major difficulty, i.e., how to choose the k initial clusters. k-means centroids true centroids • Rek-means: The prefix "Re" stands for rectangle. • AD-Test: a validating test is performed using the Anderson-Darling Statistic. • Experiments: several tests show the effectiveness of Rek-means in correctly distinguishing boats that are very close.