Share Email Print
cover

Proceedings Paper

A local distribution based spatial clustering algorithm
Author(s): Min Deng; Qiliang Liu; Guangqiang Li; Tao Cheng
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

Spatial clustering is an important means for spatial data mining and spatial analysis, and it can be used to discover the potential spatial association rules and outliers among the spatial data. Most existing spatial clustering algorithms only utilize the spatial distance or local density to find the spatial clusters in a spatial database, without taking the spatial local distribution characters into account, so that the clustered results are unreasonable in many cases. To overcome such limitations, this paper develops a new indicator (i.e. local median angle) to measure the local distribution at first, and further proposes a new algorithm, called local distribution based spatial clustering algorithm (LDBSC in abbreviation). In the process of spatial clustering, a series of recursive search are implemented for all the entities so that those entities with its local median angle being very close or equal are clustered. In this way, all the spatial entities in the spatial database can be automatically divided into some clusters. Finally, two tests are implemented to demonstrate that the method proposed in this paper is more prominent than DBSCAN, as well as that it is very robust and feasible, and can be used to find the clusters with different shapes.

Paper Details

Date Published: 30 October 2009
PDF: 10 pages
Proc. SPIE 7495, MIPPR 2009: Automatic Target Recognition and Image Analysis, 74950D (30 October 2009); doi: 10.1117/12.833567
Show Author Affiliations
Min Deng, Central South Univ. (China)
Qiliang Liu, Central South Univ. (China)
Guangqiang Li, Central South Univ. (China)
Tao Cheng, Univ. College London (United Kingdom)


Published in SPIE Proceedings Vol. 7495:
MIPPR 2009: Automatic Target Recognition and Image Analysis

© SPIE. Terms of Use
Back to Top