Author(s)

ZhiHua Liu

Affiliation(s)

JiangXi University of Science and Technology, Ganzhou 341000, Jiangxi, China.

Corresponding Author

ZhiHua Liu

ABSTRACT

Spatial clustering is the basis of pattern recognition, which is of great significance to map generalization and map updating. Aiming at the problem that many clustering methods can not effectively use the building topology information and it is difficult to deal with high-dimensional data, an attention-driven fusion clustering method is proposed. The model consists of four modules : autoencoder ( AE ), graph convolutional neural network ( GCN ), attention-driven fusion module ( AFGCN-H ) and self-supervised module. AE and GCN are used to extract features from the original data describing the characteristics of buildings and the topological information of the spatial relationship of buildings, respectively. The AFGCN-H adaptively fuses the learning representations of different layers of the two modules. The self-supervised module will optimize the clustering label allocation through corresponding losses, adjust the network parameters, learn the features suitable for clustering, and improve the accuracy of clustering. This paper uses field data sets for clustering analysis, and compares the clustering effects of traditional k-means algorithm, DBSCAN algorithm and MST algorithm. The experimental results show that the proposed method is superior to the traditional spatial clustering algorithm in the results of vector building clustering analysis.

KEYWORDS

Map generalization; Spatial clustering; Building group; Deep clustering

CITE THIS PAPER

ZhiHua Liu. An attention-driven building clustering approach taking shape features into account. Eurasia Journal of Science and Technology. 2025, 7(3): 9-15. DOI: https://doi.org/10.61784/ejst3081.

REFERENCES

[1] Tardioli G, Kerrigan R, Oates M, et al. Identification of representative buildings and building groups in urban datasets using a novel pre-processing, classification, clustering and predictive modelling approach. Building and Environment, 2018, 140: 90–106.

[2] Vinod H D. Integer programming and the theory of grouping. Journal of the American Statistical Association, 1969, 64(326): 506–519.

[3] Guha S, Rastogi R, Shim K. CURE: An efficient clustering algorithm for large databases. ACM Sigmod Record, 1998, 27(2): 73–84.

[4] Zhu Y, Jiang Y. Optimization and parallelization of BIRCH clustering algorithm. Computer Engineering and Design, 2007, 28(18): 4345–4346, 4369.

[5] Li S, Zhang H, Wang L, et al. Meteorological drought risk zoning and analysis of winter wheat in Henan Province based on grid and fuzzy clustering. Journal of Henan Agricultural Sciences, 2020, 49(11): 172–180.

[6] Tang G, Xing C, Zhu L, et al. Rapid classification of moving surface fitting filter algorithm using hierarchical clustering. Engineering of Surveying and Mapping, 2021, 30(3): 32–40.

[7] Deng M, Liu Q, Cheng T, et al. An adaptive spatial clustering algorithm based on Delaunay triangulation. Computers, Environment and Urban Systems, 2011, 35(4): 320–332.

[8] Cai J. MST clustering of urban buildings considering Gestalt principles. Chang'an University, 2018. (Doctoral dissertation)

[9] Gong X, Fang Y. Comparative study on adaptability of settlement clustering analysis algorithms. Engineering of Surveying and Mapping, 2020, 29(5): 1–7.

[10] Pilehforooshha P, Karimi M. A local adaptive density-based algorithm for clustering polygonal buildings in urban block polygons. Geocarto International, 2020, 35(2): 141–167.

[11] Van der Maaten L, Hinton G. Visualizing data using t-SNE. Journal of Machine Learning Research, 2008, 9(Nov): 2579–2605.

[12] Basaraner M, Cetinkaya S. Performance of shape indices and classification schemes for characterising perceptual shape complexity of building footprints in GIS. International Journal of Geographical Information Science, 2017, 31(10): 1952–1977.

[13] Ai T, Zhang X, Zhou Q, et al. A vector field model to handle the displacement of multiple conflicts in building generalization. International Journal of Geographical Information Science, 2015, 29(8): 1310–1331.

[14] Meng N, Feng J, Jia Y. Comparative analysis of areal settlement clustering methods. Geomatics and Information Science of Wuhan University, 2022, 47(10): 1–5.