As many kinds of Geographic Information System (GIS) data are used in the public and private sectors, the interoperability of multisource GIS data is increasingly important. However, it is difficult to integrate buildings in multisource GIS data due to differences between management agencies and geometric systems. In this study, a hierarchical iterative closest point (ICP) matching method is proposed to detect changed objects or nonidentical objects in multisource building GIS data. First, corresponding blocks between two sets of GIS data were determined. Then, building polygons located in corresponding block pairs were registered using the ICP algorithm. Based on the similarity of buildings, matching pairs were classified into two categories, matching and nonmatching, using automatic Otsu thresholding. Building pairs with ambiguous similarity based on a threshold value were processed by secondary ICP registration of individual buildings. Three kinds of criteria for re-search candidates were tested to evaluate the performance of the proposed hierarchical ICP matching method. The proposed method showed high-quality geometric registration and improved the automatic update accuracy of changed objects.
[1]
Tinghua Ai,et al.
Pattern classification approaches to matching building polygons at multiple scales
,
2012
.
[2]
Hao Yanling,et al.
Research on areal feature matching algorithm based on spatial similarity
,
2008,
2008 Chinese Control and Decision Conference.
[3]
N. Otsu.
A threshold selection method from gray level histograms
,
1979
.
[4]
Paul J. Besl,et al.
A Method for Registration of 3-D Shapes
,
1992,
IEEE Trans. Pattern Anal. Mach. Intell..
[5]
Gérard G. Medioni,et al.
Object modelling by registration of multiple range images
,
1992,
Image Vis. Comput..
[6]
Ashok Samal,et al.
A feature-based approach to conflation of geospatial sources
,
2004,
Int. J. Geogr. Inf. Sci..
[7]
Kiyun Yu,et al.
Automatic Detection of the Updating Object by Areal Feature Matching Based on Shape Similarity
,
2012
.