Global Mapper Pro

Ground Point Classification steep forest

Hello,


I am working with a very accurate point cloud with 268,000,000 points. The area is a steep mountain with forest. The maximum slope is 80-90% and the height difference of the whole point cloud is 500 meters. I need a DTM from this data. I used the tool "Automatic Ground Classification" and tested different settings on a small section of the terrain. This setting which I finally used is attached in a screenshot. However, despite this I have many outliers. These are mainly caused by tree trunks, but also by tree tops. I can reduce the number of outliers with the Automatic Noise Classification, but the remaining ones cause the DTM to be unsatisfactory. I now see as a last resort to manually classify the remaining outliers. Does anyone have a suggested solution to my problem? Would the Point Cloud Segmentation tool help? I have tried it so that the trees are segmented. But I can't get a suitable result with the setting. 



Automatic Ground Classification


Point Cloud Segmentation


Greetings 

Sandro

Best Answer

  • bmg_mike
    bmg_mike Global Mapper Guru
    Accepted Answer

    Hi Sandro,

    Automatically classifying ground in regions of extreme slope is very challenging. If there aren't any building features in your area of interest, reducing the 'Maximum Height Delta' in the automatic ground classifier could help some.

    That said, I expect some combination of Segmentation parameters might be the best approach to get large chunks of ground in a segment. If you color by segment and check the 'Select Lidar Segments' option, you can quickly use the Digitizer Tool to select entire segments and then classify all of the points as ground.

    If any manual cleanup is needed, typically the path profile tool is the easiest way to edit the classifications.

    Thanks,

    Mike

    Global Mapper Guru

Answers

  • Hello Mike,

    unfortunately, I do not manage to segment the terrain sufficiently. No matter which setting I try, nothing suitable comes up. However, after the automatic classification, I manually digitized the entire terrain using the Path Profile tool. While there are still some outliers, it looks much better. Thanks!


    Greetings 

    Sandro

Sign In or Register to comment.