Abstract: Recognition and extraction of a dual-carriageway road is a key task in road network generalization for large scale maps. Based on the analysis of dual-carriageway road recognition, this paper presents a new distance metric algorithm, facing project distance, which can measure line-line distance. A dual-carriageway road candidate set is extracted through buffering, dual-carriageway roads can be recognized accurately using constraint parameters characterised by the facing project distance. A comparison of the performance of this algorithm and existing distance metric algorithms, show that facing project distance accurately reflects the spatial proximity of a pair of lines exptessing dual-carriageway roads, the proposed algorithm can correctly recognize the dual-carriageway roads.