An automatic hybrid method for ground filtering in mobile laser scanning data of various types of roadway environments

Mobile laser scanning (MLS) has attracted increasing interest in recent years for precise mapping of roads and highways. MLS data is used for road and road-side inventory analysis, where ground filtering serves as initial but important step. Therefore, we propose a hybrid ground filtering method, with four well-designed steps: point cloud partitioning, interquartile range-based ground segment identification, ground filtering in the labelled lowest clusters, and adaptive threshold-based ground point retrieval. The proposed method was tested and validated comprehensively using eight MLS datasets from all types of roadway environment. The ground points were filtered out in these data sets at average total error and kappa coefficient of 1.73% and 96.36%, respectively. The proposed method performs satisfactorily in the challenging specific cases of terrain features and achieves significant improvement in comparison with several state-of-the-art methods. The proposed method has potential for wider use in industry as it is straightforward and computationally efficient.