Abstract: A new method for extracting bridges from airborne LiDAR data based on terrain features is proposed.A filtering-recognizing strategy is implemented.Firstly,an initial extraction is implemented on non-ground laser footprints that are labelled by a prior filtering algorithm with strict criterions.Then,an Alpha-shapes algorithm is introduced to segment the initial extraction results.Among the segments,those who belong to the main bridge body are identified by using an area threshold.Finally,a region growing algorithm based on terrain features is proposed to accurately extract the laser footprints on the bridge-ground connecting sections.The experiments are conducted to demonstrate the efficiency of the proposed method both on water and land bridges dataset.