Page Comparison

This function allows the user to detect bunches of points that are close to each other and regroup them into clusters of points.

Normally in a point cloud there is something that links the points together -  this could be the ground, the floor or the walls in the case of a building scan. If we can isolate these points from the point cloud it will be possible to detect clusters of points. This means there are a few steps to follow before being able to detect clusters. The user must first detect the ground points (or the floor points – or the wall points), send the points to a specific class and hide these points by unchecking them in the class list. Once these points are hidden, there should be unconnected clusters of points. These are the objects that will be detected with this function.

Hiding the ground emphasizes groups of points. For example; cars, road signs, utility poles, trees, people, buildings are all objects that can be singled out once the ground is hidden.

The documentation for this tool is split into separate categories:

• Cluster library: deciding where to place the cluster library
• Detection parameters: explanations for the different detection parameters
• Using the clusterization tool: how to correctly run a cluster detection
• Cluster categories and filters: how to filter clusters properly and create cluster categories
• Classifying clusters by type: using the cluster tool to classify clusters