Multi-Density Clustering Algorithm Based on Grid and Boundary

In traditional grid clustering algorithms, the cluster results are just consisted of dense grids so that the clustering quality is low, while these algorithms are unable to cluster the multi-density datasets. In this paper, we propose a clustering algorithm based on grid and boundary over multi-density datasets. In order to describe the data distribution, boundary grid is introduced and checked by the difference of density of adjacency grids. The final clusters are defined as the set of internal grids and boundary grids, which internal grids encircled by boundary grids. Experimental results indicate that our algorithm is able to cluster the multi-density datasets and has a higher clustering quality.