How can I automatically separate terrain points from the rest of the point cloud?

The software primarily uses the Selection by CSF (Cloth Simulation Filter) algorithm to efficiently identify the terrain surface. This process simulates the "draping" of a virtual cloth over the inverted point cloud; points that remain at a minimum vertical distance from this cloth are classified as ground, while the rest are identified as non-terrain. It is a very versatile tool that allows you to configure parameters such as the cloth grid resolution and the number of iterations to adapt to both flat terrain in urban areas and mountainous zones with dense vegetation.

As an advanced alternative, the user can employ the AI Classification command, which uses Artificial Intelligence models specifically trained to recognize environmental features. This method analyzes the geometry of the point cloud and automatically associates the detected points with standard LAS categories, allowing the "Ground" class to be intelligently isolated along with other elements such as buildings or vegetation.

Related videos