CLUSTERING WITH OBSTACLES IN SPATIAL DATABASES

Clustering large spatial databases is an important problem, which tries to find the densely populated regions in a spatial area to be used in data mining, knowledge discovery, or efficient information retrieval. However most algorithms have ignored the fact that physical obstacles such as rivers, lakes, and highways exist in the real world and could thus affect the result of the clustering. In this paper, we propose CPO, an efficient clustering technique to solve the problem of clustering in the presence of obstacles. The proposed algorithm divides the spatial area into rectangular cells. Each cell is associated with statistical information used to label the cell as dense or non-dense. It also labels each cell as obstructed (i.e. intersects any obstacle) or non-obstructed. For each obstructed cell, the algorithm finds a number of non-obstructed sub-cells. Then it finds the dense regions of non-obstructed cells or sub-cells by a breadth-first search as the required clusters with a center to each region. 1