Cluster Detection

Menu: Analyze / Clusters

🚀 Available in VisionLidar Premium and Ultimate versions only.

1. How Cluster Detection Works

In a standard point cloud, most points are connected by common surfaces (e.g., ground, floors, or walls in a building scan). To isolate clusters effectively, follow these steps:

1. Detect and classify non-cluster points (e.g., ground, floors, or walls).

2. Send these points to a specific class and hide them by unchecking the class in the Explorer.

3. Once these points are hidden, unconnected groups of points (clusters) will become more apparent.

4. Run the cluster detection function to identify and group these isolated clusters.

✅ Examples of objects detected as clusters:

• Cars, road signs, utility poles, trees, people, buildings

By removing the ground or other continuous surfaces, these objects stand out as separate point clusters, making them easier to detect and classify.

2. Cluster Detection Documentation

The Clusters tool documentation is divided into several sections:

📌 Cluster Library

• Defines where detected clusters are stored and managed.

📌 Detection Parameters

• Explains the available settings and thresholds for cluster detection.

📌 Using the Clusterization Tool

• Guides users through the correct workflow for running cluster detection.

📌 Cluster Categories and Filters

• Describes how to filter clusters and create custom categories.

📌 Classifying Clusters by Type

• Explains how to assign detected clusters to specific object types.