Abstract
Abstract. Filtering is one of the core post-processing steps of airborne LiDAR point cloud data. It is difficult for traditional mathematical morphology filtering algorithms to preserve sudden terrain features, especially when using larger filtering windows. In this paper, an improved progressive mathematical morphology filtering algorithm is proposed to solve the problem which is difficult to filter out a large area of non-ground points effectively and causes omission filtering on prominent topographic features. First the elevation information of point cloud data is meshed, and then the opening operation (erosion and dilation) is performed. By improving the mathematical formula of window size, the window size and the corresponding elevation difference threshold are iterated continuously. Within each corresponding filtering window, objects that are larger than the size of the structural element window are retained, and objects smaller than the size of the structural element window are filtered. Fourteen samples provided by ISPRS committee were selected to test the performance of the proposed method. Experimental results show that the improved method can effectively filter out most of the non-ground points, and this method can achieve great results not only in urban flat areas, but also in the mountains. Compared with the traditional filtering methods, the filter performance of the new method proposed in this paper has been greatly improved. The method in this paper obtains the lower errors and retains the complex topographic features.
Highlights
By improving the size of the filtering window, an improved progressive mathematical morphology filtering algorithm is proposed to filter out larger area of object points and reduce the leakage phenomenon
By judging that the height difference of a certain grid point before and after the opening operation is smaller than the height difference threshold set by the current iteration, it is recognized as a ground point, otherwise it is an object point
In view of the traditional mathematical morphology filtering algorithm, because of the small window size, the height difference between the two iterations is relatively small that resulting in some non-ground points being difficult to filter out and missing the ground point existing in the existing window size
Summary
LIDAR (Light Detection and Ranging) filtering is a challenging task, especially for area with high relief or hybrid geographic features (Wan et al, 2018). Reference (Vosselman, 2000) proposed a slope-based filter that identifies ground data by comparing slopes between a LIDAR point and its neighbors. The training datasets have to include all types of ground measurements in a study area to achieve good results. Both omission and commission errors were large when this method was applied to vegetated mountain areas with a considerable slope variation. Morphological opening operation is performed iteratively to remove object points by gradually increasing the filter window size and the elevation difference thresholds. By improving the size of the filtering window, an improved progressive mathematical morphology filtering algorithm is proposed to filter out larger area of object points and reduce the leakage phenomenon. The proposed algorithm is validated by the public test data published by ISPRS, and the effectiveness of the proposed improved method is verified, compared with the filtering performance of other algorithms
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
