Voxel Cloud Display
Voxelization (enabling the voxel processing) provides an alternative to using the Cloud Display Control to help visualize large point cloud files by providing a 3D grid or volume based sub-sampling option. It also provides an alternative to vectors for displaying cloud to object deviations or other Surface Analysis displays
How it works is you specify a grid size or volume and then only the point closest to the average for a specified volume, which we call a “voxel”, is displayed. Displaying a cloud in this way creates a roughly uniform density display of cloud data cross the entire scan region, and the size of the voxel point (displayed as a screen oriented disc or blotch) can be adjusted as a percentage of the volume used in its computation.
Enabling Voxel Display for an Existing Cloud
The Voxel cloud display can be enabled either through a cloud’s properties or as an alternative vector display option from within a Cloud to Objects relationship
The Voxel Setting dialog provides the ability to adjust the display as needed with the following controls:
-
Voxel Size. This entry field defines the length of a single side of the 3d volume used to identify each voxel. Much like defining a grid for what data to keep, this control defines the data density. The Auto Detect button provides a convenient way to compute an estimated best size to balance detail and screen performance.
-
Minimum Point Count Per Voxel. This value defines the minimum number of cloud points required in a volume in order to compute a voxel. If there are fewer cloud points than the specified number the voxel will be ignored. This provides a means to ignore outlier points.
Traditional vectors reports consider cloud point deviations no different than points. You can sub-sample and you can ignore edge projections but its still just considering the points directly. Voxels are different. They are volume oriented. In each voxel volume the cloud points are processed and only the single cloud point closest to the average in that volume is considered. This dramatically down samples the cloud statistics and distributes the analysis much more equally across the part. In addition there is some built in outlier rejection. If a considered volume only has 1 or 2 points it, it will be ignored, which is ideal for ignoring noise from a scanner. This is an adjustable threshold (Min Point Count Per Voxel) that defaults to 3. If you set this value to 6 for example there has to be 6 points in that volume or no voxel will be built in the zone. The advantage in this method is that it is not just smoothing by a distance but instead it preserves the actual scanned data. If there is a blemish in the part then removing outliers by a distance threshold may hide those bad places on the part. Voxels will not.
If you change the Min Point Count Per Voxel to 1 and turn the option to ignore edge projections off you should see the Total and Valid point counts are equal
This statistic is basically telling you how many voxel volume’s were processed and ignored. The Voxel Size defines the volume considered so with larger point clouds you will want a larger volume. For quick graphics displays it seems that keeping the total displayed number of voxels to roughly 200K is a good rule of thumb for a reasonable laptop. The Auto Detect button tries to do that for you but is generally a little conservative.
Voxel Display Settings
The Voxel Display Settings are used to control the color and size of the blotch used to display the voxel.
-
Enable Fast Rendering. This check box provides a method to disable blotch display and show voxels as regular cloud points only.
-
Voxel Rendering Diameter. Voxels are displayed as blotches or round discs oriented toward the screen, unless Fast Render mode is enabled. The size of these blotches is controlled as a percentage of the Voxel Size. This value can be adjusted to fill gaps or more clearly separate individual voxels. For a smooth colored display a slightly larger value is best such as 125%.
-
Voxel Render Border. The depth and 3D nature of a voxel dis- play is dependent upon a black border around the voxel. The thickness of this border can be adjusted if the voxels overlap greatly and the color is overly black.
Using Voxel Colors for Deviation and Analysis
Surface Analysis Mode
Each voxel point can be colorized to display different information us- ing the following settings:
-
None. Displays using the cloud color. This color is set when the voxel display is enabled so if the cloud color is changed after voxel display is set then the voxels will need to be recomputed for the color to change.
-
Relationship. In this mode, each voxel is colorized based upon the distance information in the relationship. A single cloud to object relationship needs to be defined using this cloud for this mode to be available. A single relationship can only control the colorization of a single cloud, and if a second relationship is added this mode will no longer be available.
-
Normals. Displays the XYZ direction vector for the voxel in RGB color based upon the working frame.
-
Curvature. This setting will colorize the voxels based upon the depth or extent of the data within each voxel. Changes in the depth distribution of the voxel relative to a fit plane for that same data can help identify transition zones.
-
Deviation RMS, Max, Avg. Provide a way to display the voxel colorization based upon each voxel’s statistics such as: RMS, Max and Average values independently. The extents for each voxel can then be compared to the surrounding voxels.
-
Pseudo Surface. This option was added to help visualize the cloud data in 3D. Each voxel point is oriented based upon a plane fit using the surrounding voxel points and colorized us- ing surface lighting.
-
Cloud Comparison Deviation. This mode is only available in a cloud generated from a cloud to cloud query. In this case the voxel colors display the deviation results.
The voxel processing takes place following the scan data arriving in SA so you will see a scan pass line appear first and then be converted to a surface voxel as you scan.
Basic Cloud Display Controls & Part Comparison In SpatialAnalyzer -
Simple Surface Deviation Display Process
To visualize the deviations between a cloud and a CAD model using Voxels follow these steps:
-
Import the CAD Model, scan the part and align to it. Then dis- connect from the instrument.
-
Build a Cloud to Objects relationship between the cloud and the CAD model.
-
Open the Relationship properties and enable the Voxel Cloud Display. Select Voxelized Cloud Settings to adjust coloring or tolerance settings
-
To add callouts displaying voxel deviation at particular locations, right click on the relationship and select Add Voxel Callout
Voxel callouts can be added through the right-click properties of a Cloud to Objects relationship or through the standard Callout drop down list. When you add a voxel callout you select a zone of voxels depending on how closely you are zoom in or out. The callout will by default identify the highest and or lowest voxel value for that region assuming all the voxel deviations are the same sign. If your selection bridges a zone where there are positive and negative values in that selection the callout will also display both the high and low values. The leader line will attach to the specific voxel being reported if selected directly or include a region based upon the zoom extent at the time the callout was added.