CloudCompare forum

Hello Everyone, and thank you for giving me the opportunity to join this Forum.
I am working whith the RANSAC function in order to identify plans in a point cloud of a room (inner plans). After entering the different parameters (minimum supported points per primitive, max. distance to primitive, etc.), some plans are identified and created. However, those plans extend well beyond the surface of point clouds to which they correspond. I take the example of an inner corner; the primitives of the two plans extend some metters beyond the corner creating a X instead of a L. And same happens with the openings (doors, big holes, etc.).
So, my question: is there any method to delimit the area of the plans to the boundaries of the area to which they belong?
Many thanks for your kind feedback!

Apart from manually segmenting the cloud first, I don't see how to do it... The RANSAC algorithm picks points in a RANdom fashion by definition, and doesn't consider the distances between points (or not that much at least).

Thank you very much for your kind feedback Daniel,
So, the plans found in RANSAC are always limited as a square, right?
I mean, if there is a wall whose projection in elevation is not a square, like the picture attached, the RANSAC function creates a square plan which containts all the alligned points (total height and total widht) but it doesn't fit to the sloping edges. Do you know any method or any software that could help to delimitate the plan with the actual outline?
Thank you very much in advance.

In CloudCompare, once the Ransac SD plugin has detected planes, the plugin also creates individual clouds that correspond to each plane (with the plane entity attached to it). You can select this cloud and fit a 'Facet' (Tools > Fit > 2D Polygon (Facet)). CC will try to detect the contour and it can even extract a concave contour (with you set a small enough 'max edge length'). This 'Facet' entity is composed of a contour (polyline) and a 2D polygon as mesh. This might be helpful to you?

Yes, that is exactly what I was looking for!
Thank you very much!!