A hierarchical spatial clustering algorithm based on field theory

In this paper, a hierarchical spatial clustering algorithm based on field theory (HSCBFT in abbreviation) is proposed. The field theory of spatial data is firstly employed to describe the interaction among spatial entities. Then, the agglomerative strategy is utilized to find clusters at different levels. Two experiments are preformed to illustrate three advantages of our algorithm. i) It can commendably meet the requirement that clustered entities are close to each other and similar in thematic attribute; ii) It can also discovery clusters with arbitrary shape and is robust to outliers; iii) It needs to input fewer parameters.