The point cloud data in the civil bridge construction scene includes a large number of vegetation, ground, and bridge construction point clouds. The extraction integrity of bridge buildings is still an issue for the existing filtering algorithms. This study proposes a bridge point cloud extraction algorithm based on combined filtering. First, the proposed algorithm applies the dispersion method to coarsely filter the vegetation by the feature of dispersion of vegetation point cloud distribution. Second, the radius filtering algorithm is improved to finely filter the residual vegetation point clouds based on the idea of radius filtering and making full use of color and elevation features. Finally, the ground point cloud is filtered using the normal filtering method. The experimental results demonstrate that the proposed algorithm for extracting bridge point clouds has a 99.3% integrity rate and 0.73% error rate. When compared to existing filtering algorithms, the proposed algorithm extracts the bridge point cloud more completely and accurately.