# Point cloud editing

> Source: https://topodrone.com/knowledge/topodrone-slam-100/point-cloud-editing/
> Section: TOPODRONE SLAM 100
> Updated: 2026-10-02

**Noise removal**
This function removes noise using the Statistical Outlier Removal algorithm.  
Select the point cloud to process and set the number of neighborhood points and the standard deviation multiple. The parameters are as follows.

![Point cloud editing](https://topodrone.com/knowledge/i/91244c736ab5e446.png)

Noise

Neighborhood points: the number of neighboring points used to calculate the mean distance and standard deviation for each point.

Standard deviation multiple: the value by which the standard deviation is multiplied.

**Framing**

This function splits the point cloud into tiles.  
Select the framing method (scale bar or fixed size), prefix, division scale, tile size, buffer range, etc. Then click 'framing' to process the data.

![Point cloud editing](https://topodrone.com/knowledge/i/1de3a74f5cd5d396.png)

Splitting into tiles

**Cloud registration**

This function registers several point clouds together.  
Before registration, add the reference point cloud and the unregistered point cloud to the display window. Registration consists of several steps:

Add the reference point cloud and the point cloud to the display window.

Select the reference and unregistered point clouds at the same time, then click 'Registration'.

Pick point pairs in the reference and unregistered point clouds. The corresponding points must be picked in the same order.

Pick at least 3 pairs of corresponding points in the reference and unregistered point clouds, in the same order.

Adjust the registration (ICP) parameters; when the RMS registration error meets the accuracy requirements, click 'convert' to complete the transformation.

![Point cloud editing](https://topodrone.com/knowledge/i/dfea820b2085c10b.png)

Cloud registration.

The ICP parameters are as follows:

Grid size: the point cloud grid size.

Number of iterations: the number of ICP iterations, usually 20.

Distance threshold: the maximum distance between corresponding points. If the distance to a candidate match is greater than the threshold, it is not used in the calculation.

Iterative distance: the difference between the distances calculated before and after an iteration; if it is less than this value, the iterations stop.

**Clipping**

This function clips the point cloud to a boundary.  
Select the data to clip, set the output method, import the clipping boundary (vector files in shp, dxf, fmb, kml formats are supported) and set the buffer range.

![Point cloud editing](https://topodrone.com/knowledge/i/ce96b00426e2ea2d.png)

Clipping

**Result folder structure**

![Point cloud editing](https://topodrone.com/knowledge/i/1d353a0a5fa2552e.png)

Result folder

- Clip: clipped point cloud data
- Denoise: point cloud data after noise removal
- Dimages: undistorted single images
- Filter: point cloud data after removing moving objects
- GCP: absolutely oriented odometry and point cloud
- Odometer: odometry data, where HF\_odometry.txt is high-frequency odometry, LF\_odometry.txt is sparse odometry and optimized\_odometry.txt is optimized odometry
- Optimizer: point cloud data after optimization
- Pano: panoramas
- Pos: image POS data, where camera\_pos.txt is the image POS file, camera\_trajectory.txt is the camera trajectory and lidar\_trajectory.txt is the lidar trajectory
- Register: point cloud data after registration
- Subdiv: point cloud data after framing
- Temp: temporary project folder with project information, raw point cloud data and the log. If users have problems, please send the log to the technical engineer
- Texture: point cloud data after colorization
- .sprj: project file