Lesson 10c2: Exploring and Classifying LiDAR Data in Global Mapper

Video Transcript
Download Video
Right-click and save to download

Detailed Description

This lesson will cover additional LiDAR point cloud processing and analysis features in Global Mapper.


Date Taken:

Length: 00:24:17

Location Taken: US


Lesson 10c2: Exploring and Classifying LiDAR Data in Global Mapper. Lesson 10c1 introduced the basics of interacting with lidar point cloud data in Global Mapper.

This lesson will cover additional lidar point cloud processing and analysis features in Global Mapper. If necessary, please review lesson 10c1 for guidance with importing LAS files, filtering, and visualizing lidar point cloud in Global Mapper.

By the end of this lesson, you will be able to view LAS classes, select points to reclassify using polygon selection and class filtering, and edit LAS classes with manual and automatic classification methods. In this tutorial we are using uncompressed LAS data, but The National Map also offers compressed LAZ data, which is compatible with Global Mapper as well.

If you’re interested in learning more about using USGS lidar data in other software packages, additional videos show how to use lidar data in ArcGIS Pro and LP360.

For this lesson, we will be using the same six tiles of LAS data from Lesson 10c1. If you have not done so already, please download the USGS_LPC_CO_SoPlatteRiver_Data_for_Lessons.zip file from our FTP site at : ftp://rockyftp.cr.usgs.gov/Training_Data/.

After you’ve downloaded the lidar data, extract the zip file into a folder on your local computer to use during the lesson.

The National Map has a download client where you can find USGS products including elevation data such as lidar point clouds at https://viewer.nationalmap.gov/basic/. If you are interested in learning more about downloading products on The National Map, be sure to check out our training videos located at http://www.usgs.gov/NGPvideos.

This exercise specifically uses Global Mapper version 20.0.0 with the lidar module installed. However, any version that is version 18.0 or newer should suffice for this tutorial.

Launch Global Mapper.

Make sure that your ‘LIDAR’ license is active before proceeding. Under the ‘Help’ menu, click on the ‘License Manager….’ In the next screen make sure the ‘LIDAR’ module is toggled on, then click ‘OK’. 

Note that Global Mapper preserves settings from previous use sessions, so the defaults in your interface may differ slightly from what is shown in this tutorial.

From the splash page, select “Open Data Files.” In the ‘Open’ window, browse to the location of the lidar point cloud LAS files that were downloaded from the FTP site. Click on the file named ‘USGS_LPC_CO_SoPlatteRiver_Lot5_2013_13SDD483396_LAS_2015’ and click ‘Open’.

We will load the LAS tile in its entirety. Therefore, when the ‘Lidar Load Options’ window appears, make sure all point data are loaded by clicking the ‘Select All’ button for point classifications and return types. Then click ‘OK’.

Note that if you have used the ‘Lidar Filter’ tool in previous sessions, you may encounter an alert message indicating that some classes are currently disabled and will not be displayed until they are enabled. Click ‘OK’.


When the lidar data load in the view window, make sure all classes are enabled. To do this, click the ‘Filter Lidar Data’ button, then select ‘Enable All’ and ‘OK’.

By default, in the 2D view window the LAS tile is displayed with ‘Color Lidar by Elevation’ symbology in an ‘Atlas Shader’ color palette where blue corresponds to the lowest elevation, and red corresponds to the highest elevation values. Because Global Mapper preserves settings from prior sessions, if you have previously altered the symbology selections, your display may appear differently. If needed, use the drop-down menus to display the LAS tile with the settings seen here.

Let’s start by exploring our LAS tile. In the ‘Control Center’ pane on the left, right click on the name of our LAS tile and select ‘Metadata’.

The ‘Metadata’ tab provides the metadata information for the selected LAS file such as ‘LIDAR Point Count’, ‘LIDAR Point Density’, ‘LIDAR Point Spacing’, Horizontal and Vertical Datum and Units, and ‘LAS _Version’. LAS version refers to the ASPRS LAS file format specification which provides a standardized open format to share LiDAR data. This LAS tile uses LAS version 1.2.

The ‘Statistics’ tab provides summary statistics of the LAS tile point cloud. In this tab we can see the total number of point returns and the distribution of points within return class types.  This tab also summarizes the point classifications within the LAS dataset.  This LAS tile uses the classifications ‘Unclassified’, ’Ground’, ’Low Point (Noise)’, ’Road’, ’Bridge’, and ‘High Point (Noise)’. Note that the classifications contained within each dataset may vary.

Now we will visualize some of the point classifications. Close the ‘Control Center’ window and open the ‘Filter Lidar Data’ window.

In the Filter Lidar window click the ‘Disable All’ button to clear all classification selections, then select ‘Ground’ and hit ‘OK’.

If you inspect the ground point classification in the 3D view, the surface is smooth without any vegetation or surface structures. Click the ‘Show 3D View button’ to open the 3D window. Once the window opens, double click on the silver bar on the top of the view window to align it with the 2D view. Then zoom-in using the wheel of the mouse to get a good view of the tile. You can also rotate the 3D view by holding the left mouse button down while moving the cursor on the screen or pan the 3D view by clicking the center mouse wheel while moving the cursor.

Now, in the ‘Lidar Filter Settings’ select the ‘Road’ classification and hit ‘Apply.’ You will now notice a fairly even distribution of red points across the tile. In the 3D view it appears these ‘Road’ classified points are high above the ground surface.

Now turn off the ‘Ground’ classification and add the ‘High Point (Noise)’ classification to the view. The ‘High Point (Noise)’ classification appears to be ground points in areas of swath overlap, whereas the ‘Road’ classified points appear to be actual High Noise points. Next let’s visualize class 17, the ‘Bridge’ points classification. The points classified as ‘Bridge’ appear to be buildings, vegetation and roads in areas of swath overlap. From visualizing these classifications independently, we can see that there are some misclassifications within this dataset. Close the 3D View window.

While there may appear to be large mistakes in classifications, remember that these project data were delivered in LAS format version 1.2, whereas the LAS version used by the software is the current version 1.4.

The American Society for Photogrammetry and Remote Sensing or ASPRS periodically updates the LAS file format including changes to point classes. For information about the current LAS file version, please see the LAS Specification v.1.4 - R14 document found at the ASPRS link shown here. www.asprs.org/wp-content/uploads/2019/03/LAS_1_4_r14.pdf

As you can see, in LAS version 1.2 the classes highlighted in blue were reserved but not yet described. But, by LAS version 1.4, many of these previously reserved classes were defined.

For example, in LAS version 1.2 class 17 is ‘reserved’, and by version 1.4 it is defined as Bridge Deck.

However, if you look at the metadata for this project you can see that this lidar collection used class 17 to classify points within the overlap between flight lines. This explains the reason for the classification discrepancy.

Now that we have identified some misclassifications within the dataset, we can modify them. Global Mapper provides several automatic and manual classification methods to edit and classify LAS classes. When you modify the classifications in Global Mapper all changes are performed within the workspace and are not applied directly to the original file. None of the reclassifications we perform will be applied until we export a new LAS file, therefore, it’s a good idea to save the Global Mapper workspace to avoid losing your changes. Click ‘File’, ‘Save Workspace As’, then browse to the location of the lidar data you downloaded earlier and save the workspace.

Though this dataset is already classified, for the sake of this tutorial we will be using both automatic and manual classification methods to classify this point cloud. When classifying an unclassified point cloud, a typical workflow is to classify noise points first, followed by ground, and then non-ground. We will start by using manual classification to reclassify some points from the ‘Road’ class to their appropriate class which is Noise.

We’ll start by filtering our point cloud to only display the ‘Road’ class as we did previously.

Now we must select the points we want to reclassify. We will use the ‘Select by Drawing Polygon’ button on the ‘Selection’ toolbar.  

Our potential high noise points occur across the entire tile, so we will use this tool to draw a polygon over the entire tile extent in the 2D window. Left click to create vertices and draw a polygon that encompasses the area of interest. Right click to finish. When you complete your polygon, all the potential noise points will be selected.

While most of these points are indeed ‘High Noise’ points, some are closer to the ground and should be excluded before performing the reclassification. Right click on the 2D window and select ‘Advanced Selection Options’, then ‘Filter Selected Lidar Points by Elevation/Color/Class/Etc’.

This opens a new window where we can refine the selection further by classes, elevation, return type and other class attributes such as intensity. First, under ‘Classification Filtering’ click ‘Clear All’ and then select the ‘Road’ class. Next, we will add ‘Height Above Ground Filtering’ to exclude low points from this selection. Click ‘Unselect Lidar Point Below’ and define the cut off value as ‘30’ meters. Then click ‘OK’. Re-open the 3D window and review the selection.

In the 3D window we can see that now only the high points are selected, whereas the low points near ground have been omitted from this selection.

Now that we have made the appropriate section, we can complete the reclassification. The ‘Lidar (Manual Classification)’ offers quick classification buttons for Ground, Low Vegetation, Medium Vegetation, High Vegetation, Building, Wire, Water and High Noise classifications.  Click the ‘Classify -High Noise’ button to perform the reclassification.

A message will pop-up asking you to confirm that you wish to reclassify all your selected points to High Point (Noise). Click ‘Yes’.

Now, the low points below the 30-meters above ground threshold remain in the Road classification and all the high points have been removed. Close the 3D window.

If we change the filter to view ‘High Points’ we can see that our selected points were successfully reclassified as ‘High Noise Points’.

We can further clean up this classification by reclassifying the overlap ground points contained within the ‘High Noise Points’ classification.’ As we did before, use the ‘Select by Drawing Polygon’ button on the ‘Selection’ toolbar to draw a polygon over the entire tile extent in the 2D view. Right click when you’ve drawn a polygon that encompasses the area of interest. When you complete your polygon, all the points in the High Noise classification will be selected.

Let’s again use the ‘Advanced Selection Options’ to refine the classification.

Under ‘Classification Filtering’ clear all values and select class 18- High Point (Noise).  Use the ‘Height Above Ground Filtering’ window to ‘Unselect Lidar Points Above’ 30 meters, then click ‘OK’. Now only the overlap ground points in the classification are selected.

To reclassify these points, let’s explore the additional manual classification options available under the ‘Edit Lidar Point(s)’ window. This is accessed by right-clicking on the selected features in the 2D window and clicking the ‘Edit Selected Features’ option.

The ‘Edit Lidar Point(s)’ window allows you to change the Classification, Elevation, Intensity and Flags of the currently selected points. Change the Classification to ‘2 – Ground’, select the ‘Overlap’ flag and then click ‘OK’. This adds an ‘Overlap’ attribute and reclassifies the selected points as ground.

Recall that class ‘17 - Bridge’ also appeared to be swath overlap of non-ground points. We can reclassify these points as class ‘12 - Overlap’. Use the ‘Lidar Filter Settings’ menu to enable all classes. Now use the ‘Select by Drawing Polygon’ tool to draw a polygon around the entire tile. Once the polygon is complete the tile will appear red indicating that all points are selected. This may take a few seconds.

As done before, right click on the 2D window and select ‘Advanced Selection Options’, then ‘Filter Selected Lidar Points by Elevation/Color/Class/Etc…’. Under the ‘Classification Filtering’ option, make sure only class ‘17-Bridge’ is selected. Then disable all ‘Elevation Filtering’ and ‘Height above Ground Filtering’ options and click ‘OK’.

The swaths of class ’17 – Bridge’ points are now selected. Right click and select ‘Edit’. Use the ‘Edit Lidar Point(s)’ tool to reclassify these points as class ’12 – Overlap’. Then click ‘OK’.

We can now observe the statistics for our currently selected points. Right click and select ‘Analysis/Measurement’ and then ‘Display Statistics for Selected Lidar Point(s)…’.

The ‘Lidar Dataset Properties’ window that appears is similar to the ‘Statistics’ tab in the LAS metadata that we observed earlier. This window shows the return distribution and attributes of the points. Additionally, we can see that all currently selected points are classified as class ‘12 -Overlap’ indicating that our last reclassification was successful.

Now close the statistics window and clear the current selection.

We’ve now manually reclassified the ‘Road’, ‘High Noise’, and ‘Bridge’ points. We can continue to repeat the process of performing manual classifications. However, Global Mapper also offers automatic classification methods to make this process more efficient. The ‘Lidar Toolbar’ contains Auto-Classify tools for Noise, Ground and Non-ground points. These auto-classify tools use algorithms to classify lidar points and enable operators to modify parameters based on local terrain, lidar point density, point filters and spatial extent.

Since we’ve classified ‘High Noise’ and Ground points with manual classification, we can now move on to classifying Non-Ground points with the Auto-Classify option. In this classification algorithm, Non-Ground points refer to Buildings, High Vegetation and Powerlines. Open the ‘Auto-Classify Non-Ground Points’ tool.

Make sure the ‘Find and Classify Likely Building and High Vegetation Points’ option is selected. The vegetation and building classification algorithm divides the data into bins to compare points to a calculated planar surface for classification. The ‘Base Bin Size to Check for Planar Points’ parameter defaults to 3 points spacings or 3 times the native point spacing of the Lidar data.  With higher resolution lidar data using smaller ‘Base Bin’ values will increase overall processing time but can also improve classification results. Change the ‘Base Bin Size to Check for Planar Points’ parameter to 1-point spacings. You can also specify bin size using meters, but for this tutorial we will use point spacings.

The ‘Advanced Settings’ options allow you to have more control in specifying the thresholds for points to be considered vegetation or buildings. The ‘Minimum Height Above Ground’ value specifies how far above ground points need to be for consideration as vegetation or building classes. This can help remove cars and other objects from classification. Set this value to 2.5 meters.  The ‘Maximum Co-Planar Distance’, ‘Minimum Vegetation Distance’, and ‘Max Co-Planar Angle Difference’ all define the relationships of above ground points to local non-ground best-fit planes. These settings can be modified by advanced users to fine-tune the auto classification. For this classification, accept the defaults of 0.08, 0.15, and 5, respectively.

Under the ‘Powerline Classification Setup’ make sure the ‘Find and Classify Likely Powerline Points’ option is deselected. Global Mapper recommends las tiles have point densities of at least twenty points per square meter to classify powerlines effectively. This dataset’s point density is roughly six points per square meter making it a poor candidate for powerline classification.

In the ‘Select Unclassified Point Cloud(s) to Find Likely Non-Ground Points’ window, you will see that the currently loaded las tile is selected. If additional tiles were loaded in the workspace, you would have the option to select or deselect them here as well.

Several opportunities exist to subset the input data. The ‘Only Classify Lidar Points Selected in Digitizer Tool’ option allows you to run the auto-classify tool on a subset of data if it is currently selected. The ‘Specify Bounds’ button allows you to define the data extent by defining coordinates or by drawing a bounding box. Lastly, the ‘Filter Points by Elevation/Class’ button opens the ‘Filter Lidar Points’ window we used earlier in the tutorial providing additional elevation, classification, return type and other attribute filtering options.

Since our dataset already contains classes that we wish to re-classify, select the ‘Reset Existing Non-Ground Points to Unclassified at Start’ option. Click ‘OK’. This may take a few seconds to complete.

If you would like additional information about advanced settings and additional set-up options and parameters, please see https://www.bluemarblegeo.com/knowledgebase/global-mapper-19/Lidar_Module/nonground_classification.htm.

Now let’s visualize our results. Change the ‘Lidar Filter Settings’ to only select ‘Class -5 High Vegetation’ and ‘Class -6 Buildings’ and click ‘OK’. Then change the ‘Lidar Draw Mode’ to ‘Color Lidar by Classification’.

You can see the point cloud now includes a High Vegetation classification (colored in green), and a building classification (colored in orange). This auto-classification tool performed reasonably well in capturing large-roofed industrial buildings as well as smaller residential buildings. It does however, misclassify some bridges as buildings.

The newly defined high vegetation classification captures much of the tree cover throughout the project tile. This is especially easy to identify on the tree-lined streets such as seen in the lower left corner of the tile”.

However, as you can see when you zoom in around the buildings, the auto-classification tool misclassified many of the building edges as ‘vegetation’. Additionally, the tool interprets many of the solar panels and rooftop appliances on large industrial buildings as vegetation. You can refine these classifications further by altering the parameters and filters in the ‘Auto-Classify Non-Ground Points’ tool, particularly the ‘advanced threshold values’, until you get better results, or by manually classifying points as we did previously.

In order to save our reclassifications, we need to export the LAS tile. Global Mapper will only export the data currently turned on in the filter. Use the ‘Lidar Filter Settings’ tool to turn on all classifications to export the full dataset.

In the ‘Control Center’ right click on the LAS tile and select ‘Layer’ then click ‘EXPORT’. Use the drop-down menu to select the ‘Lidar LAS File’ export format and click ‘OK’.

A tool tip pops up to inform you that the LAS file will be exported with the current display projection. This will export the LAS tile using the current horizontal projection but will not save a vertical projection. Click ‘OK’. In the ‘Lidar LAS/LAZ Export Options’ window click on the ‘Vertical Coordinate System’ drop-down and select ‘NAVD88’ to match the input data. Accept the remaining default options and click ‘OK’.

Now, navigate to where you saved the training data LAS tiles and save this exported tile as ‘Lesson10c2_LAS_Edits.las’. This may take a few seconds to export.

Congratulations! You’ve finished Lesson 10c2: Exploring and Classifying LiDAR Data in Global Mapper.

In this lesson, we discussed how to view LAS class codes, select points of interest using polygon selection and class filtering, use manual and auto-classification methods to edit LAS classes and how to export completed products. If you are interested in learning more about using lidar data, please see the additional USGS lidar training videos in ArcPro, LP360 and Global Mapper at:  http://www.usgs.gov/NGPvideos.