# Statistical Outlier Removal

* Perform Statistical Outlier Removal on the dataset

1. Specify Filter as "Statistical Outlier Removal"
2. Select input class(s)
3. Select "Low Point (noise)" as output class 
4. Click "Run Filter"
5. After processing has finished click "Save Classification to Cloud" to save changes, or click "Cancel Fitler" to Cancel changes

* The example settings below would result in all input points assigned class "Created, never classified" being output to class "Low Point (noise)" 

![](/files/-M_XlhiJ3ecuYv5AuzjA)

* This filter computes statistical outlier removal, that is it classifies all points further than X \* sigma from the local centroid, where sigma is also computed locally. 
* The exact workflow of the algorithm is as follows for each point:
  * Find its nearest neighbour
  * Find the nearest neighbour's knn
  * Compute the centroid
  * compute the mean distance from the centroid, of all knns, as well as standard deviation
  * If the point's distance is beyond X \* sigma, classify as noise

### Parameters

* **Sigma Multiplier** - Sets the X of the X \* sigma parameter 
  * The lower this multiplier is, the more aggressive the filter gets. 
* There are currently 2 distance metrics, **point-to-point** and **point-to-plane**. 
  * For open field datasets, point-to-plane might do a better job, but for indoor SLAM datasets, point-to-point provided better results during testing. 
* **Neighbors** - Number of neighbouring points to search
  * Good default parameters are 32 neighbours and 4.0 sigma multiplier


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs-6.phoenixlidar.com/lidarmill-desktop/post-processing/workflow-kk/classification/statistical-outlier-removal.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.