SphereFollowing Recursive
Plugin : SimpleForestNom de classe : SF_StepSphereFollowingRecursive
Description
First from an input cloud and an input qsm unfitted points are estimated. Those points are clustered. Each cluster large enough gets processed with the Spherefollowing routine. This produces a sub qsm representing a before not fitted branch. This subqsm is attached to the input qsm.
Paramètres
Paramètres de pré-configuration (non modifiables une fois l'étape ajoutée) :
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:
------------------------------------------------------------------------------
For this step please cite in addition (section 4.5. Imperfect Point Clouds):
Paramètres de l'étape :
Recusion, see (Hackenberg et al 2014 - section 4.5. Imperfect Point Clouds):
Pre Processing:
SphereFollowing Method Optimizationable Parameters:
Cloud To Model Distance:
For the grid search evaluation the parameter set with the smallest qsm to cloud distance is chosen.
SphereFollowing Method Hyper Parameters:
First Hyper Parameters are set. Those parameters will never change during optimization.
- Uncheck to deactivate parameterization possibilities of this step. Only recommended for beginners: Activé.
- In beginner mode select point cloud quality: medium.
- You can let Simple Forest choose method parameters: Activé select if you want auto parameters.
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:
------------------------------------------------------------------------------
For this step please cite in addition (section 4.5. Imperfect Point Clouds):
Paramètres de l'étape :
Recusion, see (Hackenberg et al 2014 - section 4.5. Imperfect Point Clouds):
- All points of the input cloud with a [point to model distance] larger 0.20 (m) are extracted..
- Those unfitted points are downsclaed with a [voxel size] of 0.01 (m). .
- The downscaled points are clustered with a [clustering range] 0.03 (m). .
- Each cluster larger than a [percentage] 0.003 (%) of the input cloud is further processed. .
Pre Processing:
- To even out the distribution and speed things up the cloud is downscaled first to [voxel size] 0.010 (m). .
- Only the largest cluster will be processed with [clustering range] 0.10 (m). .
SphereFollowing Method Optimizationable Parameters:
- You can let Simple Forest choose the following method parameters: Activé select if you want [auto parameters].
- To generate a sphere each fitted circle is multiplied with the [sphere multiplier]: 2.00 ..
- We search for sphere multiplier with the following potential parameterization: 75; 100; 150; 200; 250 %..
- The sphere surface has a thickness of [sphere epsilon] 0.01 (m)..
- We search for sphere epsilon with the following potential parameterization: 75; 100; 150; 200; 250 %..
- Point on sphere surface are clustered with threshold [euclidean clustering distance] 0.03 (m) to build new circle clusters..
- We search for euclidean clustering distance with the following potential parameterization: 75; 100; 150; 200; 250 %..
Cloud To Model Distance:
For the grid search evaluation the parameter set with the smallest qsm to cloud distance is chosen.
- For the cloud to model distance we choose [distance method] SECONDMOMENTUMORDERMSAC - minimize cropped (MSAC) root squared distance .
- For MSAC and inlier methods the distance is cropped at [crop distance] 0.030 (m)..
SphereFollowing Method Hyper Parameters:
First Hyper Parameters are set. Those parameters will never change during optimization.
- [SAC Model Consensus method] Center of Mass with median distance radius .
- The [inlier distance] for the selected SAC Model Consensus method is 0.030 (m). .
- For fitting a geometry with the selected SAC Model Consensus method at minimum [minPts]: 3 points are needed..
- For the selected SAC Model Consensus method [iterations]: 100 are performed..
- The radius of the search sphere during sphereFollowing is never smaller than [min global radius] 0.010 (m). .
- The algorithm is initialized on a close to ground slice with height [initialization height]: 0.05 (m)..
Données d'entrée
Structure des données d'entrée recherchées :
Result : QSM result
...
QSM Group (Group)
internal QSM ()
Result : Input Result QSM SphereFollowing
...
Tree Group (Group)
Input Cloud QSM SphereFollowing (Item with points)
Result : QSM result
...
QSM Group (Group)
internal QSM ()
Result : Input Result QSM SphereFollowing
...
Tree Group (Group)
Input Cloud QSM SphereFollowing (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.
Hackenberg Jan, Sheppard Jonathan, Spiecker Heinrich, Disney Mathias. 2014. Highly Accurate Tree Models Derived from Terrestrial Laser Scan Data: A Method Description. 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.
Hackenberg Jan, Sheppard Jonathan, Spiecker Heinrich, Disney Mathias. 2014. Highly Accurate Tree Models Derived from Terrestrial Laser Scan Data: A Method Description. 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.