Fig. 1. Flow chart of algorithm

Fig. 2. Number of signal pulses in different filter core shapes

Fig. 3. Slope adaptation in flat areas

Fig. 4. Slope adaptation in areas with large gradient

Fig. 5. Histogram of point cloud density

Fig. 6. Point cloud density histogram with different noise rates

Fig. 7. Data distribution in the study area

Fig. 8. Original point cloud of data A

Fig. 9. Original point cloud of data B

Fig. 10. Adaptive results of spatial density of data A

Fig. 11. Adaptive results of spatial density of data B

Fig. 12. Final denoising result of data A

Fig. 13. Final denoising result of data B

Fig. 14. Comparison between signal point cloud extracted in this paper and ATL03 signal in data A

Fig. 15. Comparison between signal point cloud extracted in this paper and ATL03 signal in data B

Fig. 16. Comparison between signal point cloud extracted in this paper and ATL08 signal in data A

Fig. 17. Comparison between signal point cloud extracted in this paper and ATL08 signal in data B

Fig. 18. Partial visual analysis of data A

Fig. 19. Partial visual analysis of data B

Fig. 20. Results of data A based on circular filter kernel

Fig. 21. Result of data A based on elliptic filter kernel