Computree Documentation

Dikstra Based Tree Segmentation

Plugin : SimpleForest
Nom de classe : SF_StepSegmentationDijkstra

Description

The step takes vegetation points and seed clusters as input. All clusters are tagged with an own id and zero distance, remaining non cluster points areinitialized with infinity distance. Then a competitive dijkstra is applied. Later all points are tagged with the id from the seed cluster they are connected to.Before the routine the cloud can be scaled in z-axis to enable easier vertical growth of the dijkstra routine.If points remain not reached by the dijkstra they are alligned to the closest pre cluster tree by nearest neighbor check.

Paramètres

Paramètres de pré-configuration (non modifiables une fois l'étape ajoutée) :

  • Uncheck to deactivate parameterization possibilities of this step. Only recommended for beginners: Activé.

Paramètres de l'étape :

  • The cloud is scaled along the z axis with a [factor] of 1.0 ..

  • The cloud is downscaled with [voxel size] 0.0300 (m)..

  • Then a Dijkstra based Segmentation is performed with neighbors connection with [range] 0.0800 (m)..


Voxel size should be at minimum 2 or 3 times smaller than Dijkstra range.

Please excuse potential double citation with the step related citation for the

following general citation for SimpleForest (work on updated citable resource ongoing):





A majority of implementations are based on PCL library:





------------------------------------------------------------------------------

Données d'entrée

Structure des données d'entrée recherchées :

Result : Scene Point Cloud
    ...
        Vegetation Group (Group)
            Vegetation Cloud (Item with points)

Result : Seed Point Cloud
    ...
        Seed Group (Group)
            Seed Clusters (Item with points)

Références

Hackenberg Jan, Spiecker Heinrich, Calders Kim, Disney Mathias, Raumonen Pasi. 2015. SimpleTree - an efficient open source tool to build tree models from TLS clouds. Multidisciplinary Digital Publishing Institute. Forests.
Rusu Radu Bogdan, Cousins Steve. 2011. 3d is here: Point cloud library (pcl). IEEE. Robotics and Automation (ICRA), 2011 IEEE International Conference on.
Gorte Ben, Pfeifer Norbert. 2004. STRUCTURING LASER-SCANNED TREES USING 3D MATHEMATICAL MORPHOLOGY. TU Delft. Section of Photogrammetry and Remote Sensing.
Dijkstra Edsger W. 1959. A note on two problems in connexion with graphs. Springer. Numerische mathematik.