For documentation on the current version, please check Knowledge Base.
Ghost Cleaning
This page describes the desktop extension “Ghost Cleaning” for Mobile Mapping.
Main Toolbar > Manage > Ghost Cleaning
Concepts
Ghosts : A ghost represents a collection of lidar points that should not be present in the point cloud. The best examples are moving objects, like cars or people that are recorded in several positions and appear as erroneous points.
Ghost cleaning: The process of removing unwanted objects from the point cloud. Orbit achieves this goal automatically inside a defined area, by creating a ground surface on the fly and removing all objects above that surface inside a fence. The Ghost cleaning extension only finds, calculates and selects the points that need to be removed, the actual removal of the objects requires Consolidation
Source
Use the generic workflow for Point Cloud Based Extensions to define the source file.
The method for the range of interest is 'Area'.
Ground Reference
- By DEM parameters: The Ground reference is defined by the parameters set in the Options.
- By DEM file: A selected DEM file will define the Ground Reference height.
- By fixed ground height: A fixed height can be used as Ground Reference height.
Options
Define the parameters that will be used in the automatic detection of ghosts.
- Pixel size: The cell size of the DEM created to define the ground surface.
- Noise Thickness: The maximum z range for the points within a pixel.
- Horizontal Neighborhood: Amount of pixels around current one where Vertical Neighborhood is applied.
- Vertical Neighborhood: The maximum difference pixel height value is allowed to have with the neighboring pixel to be considered “ground”.
- Ground Margin: The minimum distance between the ground surface and a point, in order to consider that point a ghost.
- Ground offset max: The maximum distance between the ground surface and a point, in order to consider that point a ghost.
Only available when the Ground Reference is set 'By DEM file'.
- DEM file: Choose the DEM file to use as ground reference.
Only available when the Ground Reference is set 'By fixed ground height'.
- Ground height: Choose the fixed height to use as ground reference.
Process
Delegate to task manager
Allows to schedule a job to be run later by delegating it to the Task Manager.
Start Now
Start the ghost detection immediately. It will create separate .ops files for each project.
Manual Adjustment
Allows the manual editing of the points to be included in the selection, and as a result modifies the objects considered as a ghost.
By pressing “Manual Adjustment” the 3D Point Cloud Clipping tool is opened, which allows manual selection inside the point cloud.
Merge Projects
Creates a temporary point cloud file necessary when the run is consolidated. Several projects can be used at once, the result will be an .opd file stored in the run's point cloud folder. The function is envisioned to be run only one time for all the projects. To run “Merge Projects” a second time, the previous .opd file must be deleted.
Start
Start the generation of the .opd file to be used in the Consolidation extension.
View Settings
Show Fence
Choose to visualize the fence that is drawn or selected.
Ghost Cleaning Algorithm
- Define your Ground Reference which can be:
- a DEM file: select a file
- a fixed ground height: one value Ground Height
- defined DEM by parameters
- Pixel size: pixel size of the DEM to be created
- Noise Thickness: Maximum z range of the points within one pixel
- Horiz Neigh: horizontal neighborhood used to remove car roofs
- Verti Neigh: vertical neighborhood used to remove car roofs
- Divide the area into 2D cells > Pixel size
- For each cell, the ground height is derived from the DEM (file, parameters, or fixed value.
- IF a point from the point cloud > Ground Margin AND point from the point cloud < Ground offset Max THEN the point is a ghost.
Ghost Cleaning Tab
In the Ghost Cleaning Tab, all Projects are shown including the results and metadata. Standard operations like detaching and maximizing tab are available.