A Gram-Schmidt Transformation-based Hierarchical Urban Road Centerline Extraction Method

This paper proposed a novel approach for extracting road centerlines in the urban area. The proposed approach consists of three main blocks: (1) DSM data are added to the available high resolution remote sensing images (HRRSI) and the two types of data are fused by the Gram-Schmidt transformation; (2) A hierarchical road extraction method is proposed to extract the road centerlines. (3) Finally, an improved road trimming (IRT) algorithm is implemented for road centerline refining. Experimental results on four datasets show that the proposed method obtains competitively satisfactory results.