A vector tessellation method is proposed for grid pattern recognition in street networks. This study regards a street network as an independent subspace embedded in the 2D space, and subdivides street segments into linear elements with equal lengths. The characteristics of grid patterns are extracted, including directional, geometrical and topological features. To map the object space to the feature space and to build a vector field, the linear element is described as a feature vector and the eigenvalues are calculated with the neighboring elements. A grid pattern classification is realized based on a support vector machine (SVM), and the classification result is optimized based on Gestalt principles. The method was applied to the street network of Shenzhen. The experimental results show that the method effectively mines grid pattern in street networks.
|Journal||Wuhan Daxue Xuebao (Xinxi Kexue Ban) - Geomatics and Information Science of Wuhan University|
|Publication status||Published - 2018|
- Feature extraction
- Grid pattern
- Spatial tessellation
- Street network space