Monitoring and maintenance activities for bridges benefit from the use of digital twins. A digital twin can be defined as a numerical model of an existing structure enhanced with real-time data, representing the actual state of an asset. The manual preparation of the underlying numerical model is considered to be time-consuming, consequently automating the Scan-to-BIM process is of particular interest. One major challenge of automation is to segment the built asset by automatically detecting its components. This contribution proposes a methodology for bridge component detection based on a template-matching algorithm. Tests showed the accuracy of the proposed solution, even for incomplete point clouds, without requiring extensive input datasets. Compared to other solutions, this approach shows potential for adaptation to a variety of bridge designs.