3D acquisition devices usually produce unstructured point-clouds as primary output. A challenge in this context is the decomposition of the point-cloud data into known parts in order to introduce abstractions of the originally unorganized data. This information can be used for compression, recognition and reconstruction.