Problem mapping a forest for DTM

Hello all,
I’m a young pilot from Cameroon. The problem I have is that our Company took a job to realize a DTM of a forest in a swamp with à normal dji phantom 4 pro v2 camera. The problem is that after the first trial, we remarked it woulf be really complicated without a clear view of the ground.
But we don’t have any lidar camera. Please do you have any idea on how to do to solve this problem ? I Heard about carlson Precision 3d topo somewhere here… Please how to ?

Thank you in advance for any advices

1 Like

I’m probably the one that recommended Carlson. Can you PM me a link to download the point cloud? I’d like to take a look at it and run at it before I make too many recommendations.

1 Like

Hi MichaelL,
Thank you for your answer.
I launched the processing of the point cloud in DroneDeploy in xyz format.

While waiting for it to be done, here is the view only link of the first trial we made :;jwt_token=eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzUxMiJ9.eyJleHAiOjI1MzQwMjMwMDc5OSwiaWQiOiI1ZTE5ZmI1MDk2NDIyODU1N2U0YTRhMDgiLCJzY29wZSI6WyJmM2E4OGFjOTA0XzRCRjg3NDE3NjJPUEVOUElQRUxJTkUiXSwidHlwZSI6IlJlYWRPbmx5UGxhbiIsIm92ZXJsYXlfZm9sZGVyX2lkIjoiNWUxODVkZmEzMmNhZjMzN2NmYTk0NDdlIn0.8Lk7lt9tY_CUBwIWtBsxGChr0rO02pz7_IB4HK_YcAjimZPO0wrBk0DNlXCMGBWVQknYISZlIP5e9PUbZwIuVQ

Thanks for this. After looking at the model the palm covers the ground too much and I don’t think it would be possible to make a DTM with photogrammetry. Even LIDAR would have trouble getting enough in-between points to make an accurate DTM with this. Unfortunately to only way I can see getting this one done is with a total station or robotic instrument. You may be able to do it with GPS if you have an extra tall rover rod.

We’ve also got a couple of holes. Comparing Carlson and DD confirming.

Carlson P3D Topo


It’s hard to explain, but basically Carlson P3D Topo removes any points a designated deviation higher than what it considers low (ground) points across another designated horizontal measurement. A smaller cell across a larger window.

Native Point Cloud (8.2M points)

This is about a 10m x 10m window with 0.3m cells.

This is the point cloud after running a 0.3m cell and 3m window with a height tolerance of 0.3m and is the normal filter that I run on a construction site that usually gets rid of small groups of trees, cars and machines. Problem is that with a small 0.3m cell it is still has too many tops of lower canopies. It appears to have removed allot of point with 1.7M remaining, but the canopy was so dense that it is really not allot removed with how few ground points there are.

Here is an aggressive 1m x 5m x 0.3m which still leaves enough points to stitch a decent DTM, but I don’t think you could go much further without having to stitch too far across. What I would do with this result is go ahead and make a triangulated surface and turn on contours so I can easily see the remaining spikes. Unfortunately it is the opposite and the ground points are the minority so it would be allot of trimming and not enough points left to stitch to where I would be comfortable presenting.

The ground points are indicated with the magenta stakes.

I have changed the thread category due to the issues with General Discussion not showing up in the Latest filter. @Andrew_Fraser

Thank you very much MichealL.
Please how can I obtain even an approximative DTM then ?

What file format is your end goal?

We need the Autocad format please

Just sent a PM so we can coordinate. I will take another look at it tonight to see if there is a different set of filters that may work better.

1 Like