====== Clash Detection ====== 

This page describes the Orbit desktop extension "Clash Detection". 

=== Feature Extraction Pro ===
{{orbitgis:how_to_find.png?25&nolink|}} [[227:desktop:workspace:main_toolbar|Main Toolbar]] > Extensions > Clash Detection\\

{{  :227:desktop_ext:clashdetection_228.png?nolink&800  |}}

===== Concepts =====

The Clash Detection extension detects clusters of points within a volume that is defined by moving a vertical shape along a 3D  path. \\
Clusters detected within the shape are hits. There are two buffers around the shape: near hits and watch list clusters. \\

The vertical shape can be added to the Clash detection via the Shape creator extension. 


===== Source =====

Use the generic workflow for [[227:desktop_ext:generic|]] to define the source file. 

The method for the range of interest is 'Path'. 

===== Shape =====

=== Drop-down menu of available shapes ===

Choose the 2D shape from the list of default and custom-made shapes. Use the [[227:desktop_ext:shape_creator|]] to add new shapes.

=== Position Shape on Path ===

Opens the "Clash Detection" tab, and gives the possibility to adjust the position of the 2D Shape on the 3D Path. By default, the shape will be centered around the path, by drag & drop it can be re-positioned.

===== Parameters:Algorithm =====

=== Get Hits ===

Detect clusters of points inside the selected shape along the path.

=== Get near hits ===

Detect clusters of points outside of the shape but within the 'Near hits margin'.  'Watch list' buffer distance parameters is expected to be bigger than one from 'Near hits'.

=== Get watch list ===

Detect clusters of points outside of the shape + near hits but within the 'Watch list margin'. 'Watch list margin' parameter is expected to be larger than the 'Near hits margin'.

=== Ignore from the bottom ===

FIXME

=== Cluster results ===

Cluster the point cloud selection for every clash type to detect different objects. 

** Cluster points min **

The minimum number of points to be considered as a clash. The cluster is cumulative, starting with the initially detected points, new points are added to the same cluster if they are in a certain 3D distance (** Cluster distance **). If new points are detected at a larger distance from the last detected point in the cluster, they will constitute a new cluster. 

=== Interpolate roll ===

FIXME

=== Classified point cloud ===

Include/Exclude classes to be used for the generation of the point cloud selection.

==== Parameters:Output ====

=== Get selection per type ===

One ops is created for the Hits, Near hits and Watch list.

=== Get selection per cluster ===

One ops is created for every cluster.

=== Outline fit rectangle ===

FIXME

=== Get outline per cluster ===

Two outline vector files (2D and 3D) is created per cluster. 

=== Apply classification on clustering ===

FIXME