Clustering Geospatial Objects via Hidden Markov Random Fields

This paper addresses the problem of clustering objects located and correlated geographically and containing multiple attributes. For the clustering problem, it is necessary to consider both the similarities of the attributes and the spatial dependencies of the objects. A new clustering framework using hidden Markov random fields (HMRFs) and Gaussian distributions and new potential models of HMRFs for irregularly located geospatial objects are proposed in this paper. Experimental results for systematic data and two real-world data showed the availability of the proposed algorithms.