An automatic approach for detecting bridges over water from light detection and ranging (LiDAR) data based on adaptive morphological filter and skeleton extraction is presented. It is inspired by data-driven and inference-based methods in machine learning. First, the three-dimensional characteristics of LiDAR data are considered in our algorithm. We design an adaptive morphological filter to classify the data into two classes, ground points and nonground points. Second, the elevation feature is used to extract the river. In this way, the search space can be greatly reduced. Third, the river is represented as a skeleton line by the morphological thinning algorithm. This concise representation makes the proposed approach more efficient to detect bridges. Finally, we propose the shortest distance rule based on the skeleton line. The fusion of the classification map and the rule is used to detect bridges. The flexibility of the proposed method is demonstrated by experiments on several different scenes. The experimental results show that the proposed approach has good performance in detecting a bridge over water.