An Improved Density-based Spatial Clustering Algorithm Based on Key Factors of Object's Distribution

Abstract

Density-based spatial clustering algorithms can be used to filter out noise and outliers, and discover clusters of arbitrary shape, which are all relatively good algorithms. But when the problem of variable density distribution of spatial objects was taken in to consideration, the accuracy of clustering result can be largely affected by the distribution of spatial objects. Therefore, the strategy of choosing the neighborhood radius threshold between objects become the key of the algorithm. An algorithm with a neighborhood radius threshold choosing strategy that based on factors that influence the distribution of spatial objects is proposed in this paper. At the same time, it adopts quadtree indexing technology to improve the efficiency of this algorithm. Experiment shows that this algorithm can efficiently deal with the problem of clustering in spatial objects’ variable density distribution.