CUBN: A clustering algorithm based ondensity and distance

In data mining, clustering is used to discover groups and identify interesting distribution in the underlying data. Traditional clustering algorithms favor clusters with spherical shapes and similar sizes. We propose a new clustering algorithm called CUBN that integrates density-based and distance-based clustering. Firstly, CUBN finds border points by using erosion operation that is one of the basic operations in mathematical morphology, then, it clusters the border points and inner points according to the nearest distance. Our experimental results show that CUBN can identify clusters having non-spherical shapes and wide variances in size, and its computational complexity is O(n). Therefore, this algorithm facilitates the clustering of a very large data set.