Fig. 1. Gridding of three-dimensional laser point cloud. (a) Raw laser point cloud; (b) grid data interpolated from point cloud

Fig. 2. Detection of original noise areas. (a) Cluster areas of grid data; (b) original noise area

Fig. 3. Process of top-hat transform. (a) Gray scale cross-section of image; (b) result of white top-hat transform; (c) result of black top-hat transform

Fig. 4. Detection of final noise areas based on top-hat transform. (a) Cross-section of the grid of raw point cloud; (b) detection of high noise based on white top-hat transform; (c) detection of low noise based on black top-hat transform

Fig. 5. Point cloud interval of flight and scanning direction in middle and lateral areas for Strip 9 in Vaihingen. (a) Point cloud interval of scanning direction in middle areas; (b) point cloud interval of flight direction in middle areas; (c) point cloud interval of scanning direction in lateral areas; (d) point cloud interval of flight direction in lateral areas

Fig. 6. Grid data interpolated from Strip 9 in Vaihingen when L=0.4

Fig. 7. Comparison and analysis of experimental results for Strip 9 in Vaihingen and Strip 5 in Toronto. (a) Raw point cloud of Strip 9; (b) denoising result of Srtip 9 using distance-based method; (c) denoising result of Srtip 9 using statistical-based method; (d) denoising result of Srtip 9 using proposed method; (e) raw point cloud of Strip 5; (f) denoising result of Srtip 5 using distance-based method; (g) denoising result of Srtip 5 using statistical-based method; (h) denoising result of Srtip 5 us