Detection Parameters

Menu: Analyze / Cluster

The Clusterization Tool detects objects in a point cloud based on two primary parameters:

✔ Separation Distance – Defines the minimum distance between point groupings.
✔ Minimum Number of Points – Determines the smallest point group that qualifies as a cluster.

1. Detection Settings

Separation Distance

• Defines how far apart point groupings must be to be considered separate clusters.

• Higher values → Increases the likelihood of merging nearby clusters.

• Lower values → May split a single object into multiple clusters.

Minimum Number of Points

• Defines the minimum number of points required for a grouping to be considered a cluster.

• Smaller values → More objects (even those with few points) will be detected as clusters.

• Larger values → Helps eliminate false clusters caused by sparse points.

2. Downsampling for Faster Detection

Downsampling can be applied before detection to speed up processing by reducing the number of points analyzed.

• The Downsample Option retains only one point per cell, discarding others.

• The Downsample Size determines cell size:

  • Larger values → Fewer points analyzed (faster but less precise).

  • Smaller values → More points retained (slower but more accurate detection).

3. Optimizing the Detection Area

Before running cluster detection, consider segmenting the point cloud to improve performance:

✔ Use a Fence or Clipping Box → The tool will only process visible points.
✔ Hide Certain Classes (e.g., Ground) → Prevents unnecessary processing of non-cluster points (highly recommended).
✔ Create an Extraction → Isolating a smaller area reduces processing time by focusing only on relevant points.

4. Handling Small Clusters

Small Cluster to Class

• Classifies small clusters (below the minimum required points) into a specific class.

• Often used to reclassify small point groups as noise.