Gird Pattern Recognition in Road Network Using rincipal Component Analysis

The spatial pattern recognition and implicit information discovery have generated interest in the field of spatial data mining. Grid pattern is one of the most typical patterns in road networks. This paper presents a method for grid pattern recognition using principal component analysis. The method first defines shape measures and relation measures to formalize meshes in road networks. The measures are: rectangularity, convexity, consistent arrangement degree, rectangularity of mesh having largest consistent arrangement degree, convexity of mesh having largest consistent arrangement degree. Second, the principal components are created and analyzed. Finally, the method identifies grid pattern via the meshes’ membership degree of the grid derived from the first principal component. Experiments on Shenzhen road network and Wuhan road network have been conducted. The results show that it is valid in grid pattern recognition and it is a versatile method to some extent.