Building Classification

Skip to first unread message

Hristiyan Dimitrov

Oct 13, 2022, 6:43:48 AM10/13/22
to LAStools - efficient tools for LiDAR processing
Hi, I am currently doing my Master's Thesis on semantic building modelling and trying to extract buildings by classification from a photogrammetric point cloud. However, the building classification gets mixed up with the vegetation classification. Could you give me advice on what the reason for that could be? A good classification is imperative for my work, since the resulting buildings are clustered and their respective point clouds are used to generate footprints and later the entire 3D models.

My best workflow after a few weeks of trying still does not deliver satisfactory results. It has the following structure (input cloud is in ECEF WGS84 coordinates):

  1. Transformation to UTM
    las2las -i h100_l80_q80_g_o_dense_georef_clipped.las
    -o h100_l80_q80_g_o_dense_georef_clipped_utm2.las -ecef -target_utm 32N
  2. Ground estimation + height calculation
    lasground_new64 -i h100_l80_q80_g_o_dense_georef_clipped_utm.las -city
    -o h100_l80_q80_g_o_dense_ground.las -step 15 -ultra_fine -ecef -compute_height
  3. Classification
    lasclassify64 -i h100_l80_q80_g_o_dense_ground.las
    -o h100_l80_q80_g_o_dense_classified.las -ground_offset 5 -keep_overhang -wide_gutters
This leads to the following result:


The result is relatively ok, but many trees, especially in the lower left part of the cloud, are wrongly classified. Furthermore, only the building rooftops are detected correctly (even when not using the -ground_offset 5 during classification).

During most of my tests, the buildings were classified partly as vegetation. Thereby, I have played with most of the parameters in lasground and lasclassify, such as step size, input reference system, offsets etc. Here is the result that I usually get (also from the upper cloud):


Both point clouds are photogrammetric, whereas the first dataset was gathered by me. Could the input data be the reason for the surprising result? The reconstruction was conducted in COLMAP, the georeferencing in Cloud Compare. I have also tried the same workflow with the wrapper transformers inside the Feature Manipulation Engine (FME) and get the same results.

I would appreciate any help or advice. Thank you in advance!

Best regards
Hristiyan Dimitrov

Daniel Andrade

Oct 13, 2022, 9:26:16 AM10/13/22
Hi, Hristiyan.. what are the values of planarity and roughness are you using in lasclassify ?? The higher the threshold for planarity, the greater the chance of overestimating buildings.. maybe that's what is happening.


Download LAStools at
Manage your settings at
You received this message because you are subscribed to the Google Groups "LAStools - efficient tools for LiDAR processing" group.
To unsubscribe from this group and stop receiving emails from it, send an email to
To view this discussion on the web visit


Oct 21, 2022, 5:54:39 AM10/21/22
to LAStools - efficient tools for LiDAR processing
Hello Daniel,

I was away, therefore the slow response. The results that you can see utilize the default values for rooftop planarity and tree forms, which are 0.1 and 0.4 respectively. I have tweaked both parameters up and down separately, as well as simultaneously. However, this brings little to no change to the result. 

The software also works for photogrammetric point clouds, right? I am wondering because they contain no additional information, such as first and last pulse.


Daniel Andrade

Oct 25, 2022, 10:59:19 AM10/25/22
Hello, Hristyan. I'm not sure if Lastools can also be used for photogrammetric point clouds without performance loss, because of point density.

Jochen Rapidlasso

Oct 25, 2022, 11:04:14 AM10/25/22
to LAStools - efficient tools for LiDAR processing
Hi all,
LAStools love to work with photogrammetric point clouds - and you may find nothing faster than LAStools.
Some algorithms are may optimized for LiDAR point clouds - but you can try and test how the tool will work with your individual set of data.
Just give it a try.

Jochen @rapidlasso

Jorge Delgado García

Oct 25, 2022, 3:34:43 PM10/25/22
Hi Daniel,

I have been using LAStools for more than 12 years and I can tell you that almost 75% of the data clouds I process are from SfM/MVS, i.e. photogrammetric techniques. The truth is that the only difference is that finally you do not have several echoes, because nowadays thanks to dense matching methods you can obtain points with a high density, without the need for high cost systems, such as aerial or terrestrial laser scanners.

In many cases, we even combine points from different sources within the same final data set. LAStools really provides a lot of flexibility in this. As for the extraction of buildings, it would be good to have some example to test, in this case the parameterization of planarity (buildings) and roughness (vegetation) is basic and depends a little on each scene.

Greetings to the whole group,


Hristiyan Dimitrov

Oct 26, 2022, 12:56:05 PM10/26/22
to LAStools - efficient tools for LiDAR processing
Hello all,

thank you for the responses. Here is the photogrammetrically derived dataset from the first image that I have been using:
The coordinates are georeferenced to WGS84 in ECEF format.


Jorge Delgado

Oct 28, 2022, 6:53:31 PM10/28/22
to LAStools - efficient tools for LiDAR processing
Good evening, Hristyan

I am attaching a file with some ideas regarding your data. I just realized that you indicated that the area was UTM32N and I used UTM34N, but the whole process would be the same.

I hope that will be useful for you (and for the group).

Prof. Jorge Delgado
University of Jaén (Spain)
Reply all
Reply to author
0 new messages