Pixels-to-Points™: Easy Point Cloud Generation from Drone Images

Point cloud generated from 192 drone images using the Pixels-to-Points tool
A point cloud generated by EngeSat’s Laurent Martin using the new Pixels-to-Points™ tool in version 19 of the LiDAR Module. The LiDAR Module tool analyzed 192 high resolution drone images to create this high-density point cloud.

When we have a new product release like the version 19 of the LiDAR Module that comes with the Pixels-to-Points™ tool, it’s always exciting to see that feature in action for the first time outside of the Blue Marble office. Our South and Central American reseller Laurent Martin from EngeSat was quick to try the new Pixels-to-Points tool for himself using drone data collected by his peer Fabricio Pondian.

The new Pixels-to-Points tool uses the principles of photogrammetry, generating high-density point clouds from overlapping images. It’s a functionality that makes the LiDAR Module a must-have addition to the already powerful Global Mapper, especially for UAV experts.

Below, screenshots captured by Laurent illustrate the simple step-by-step process of creating a point cloud using the Pixels-to-Points tool and some basic point cloud editing using other LiDAR Module tools.

1. Loading drone images into the LiDAR Module

The collection of images loaded into the LiDAR Module must contain information that can be overlapped. The Pixels-to-Points tool analyzes the relationship between recognizable objects in adjacent images to determine the three-dimensional coordinates of the corresponding surface. In this particular example of the Pixels-to-Points process, 192 images are used.
The flight path of the UAV and the locations of each photo can be viewed over a raster image of the project site.

2. Calculating the point cloud from loaded images

192 high-resolution images are selected in this particular example. The tool will give an estimated time of completion, which depends on the size of the images and number of images.
The Calculating Cloud/Mesh dialogue displays statistics of the images as they are analyzed and stitched together by the Pixels-to-Points tool.
An alert window pops up when the process is complete.

3. Viewing the generated point cloud

A new layer of the generated point cloud is now in the control center.
A close up of the final processing result with the orthoimage.
A close up of the final result with the new point cloud generated from the 192 images.
A 3D view of the resulting point cloud.
A view of the point cloud colorized by elevation
A cross-sectional view of the point cloud using the Path Profile tool

4. Classifying the point cloud

Points can be reclassified automatically or manually using LiDAR Module tools. Here, the point cloud is reclassified as mostly ground points.

5. Creating an elevation grid and contours from the point cloud

With the point cloud layer selected, a digital terrain model can be generated by clicking the Create Elevation Grid button.
A cross-sectional view of the digital terrain model using the Path Profile tool
Contours can be generated from the digital terrain model by simply clicking the Create Contours button.

A quick and easy process

In just a few steps, Laurent was able to create a high-density point cloud from 192 images, reclassify the points, and create a Digital Terrain Model. It’s a prime example of how easy version 19 of the LiDAR Module and the new Pixels-to-Points tool are to use. Check out EngeSat’s full article on the release of LiDAR Module.

Where in the World October 2017 Answers

Name the country – Djibouti

Djibouti

 

 

 

 

 

 

 

 

 

 

 

Name the river – The St. Lawrence River

The St. Lawrence River

<

 

 

 

 

 

 

 

 

 

 

Name the island – Borneo
Borneo

 

 

 

 

 

 

 

 

 

 

 

Name the capital city – Lisbon
Lisbon

 

 

 

 

 

 

 

 

 

 

 

Name the mountain range – The Pyrenees
The Pyrenees

Estimating Property Modifications in Global Mapper

Connecting to the US NAIP high-resolution imagery.
Connecting to the US NAIP high-resolution imagery.

 

I recently purchased a house in Hallowell, Maine, where the Blue Marble Geographics office is located. Hallowell is a teeny tiny city with lots of historic homes that sit on a rather large hill overlooking the Kennebec River. One aspect of my historic fixer-upper property that needs some work is the drainage. I have decided to explore drainage solutions by estimating property modifications using Global Mapper and publicly available data.

Finding Data in Global Mapper

The first step is finding the right data. So, to start with, I use the search tool in Global Mapper to create a point feature at my address. I also change the projection to something that works for the area, such as the State Plane projection for Maine. Next, with the online data tool, I easily connect to the US NAIP high-resolution imagery.

The State of Maine GIS site, MEGIS, has a number of other helpful layers that can be added. Vector data can be downloaded as shapefiles using a web browser and can be loaded into Global Mapper by simply dragging the files into the software. Like a lot of states, Maine’s GIS site also offers web services that can be added to the list of online sources in the software. For my project, I need the outline of my individual property, so, I first download the property parcels layer for the entire city and drag the downloaded zip file onto the map to import it. I use the Digitizer to select my property and then use CTRL+C and CTRL+V to copy it to a new layer.

Using the Digitizer tool
I use the Digitizer to select my property and then use CTRL+C and CTRL+V to copy it to a new layer.

What I really need for this analysis is some high-resolution terrain data, and luckily my property is close enough to the coast to be included in the NOAA coastal LiDAR data. I use the online data source tool again to search the Digital Coast for data that matches my current map bounds.

Cleaning up LiDAR Data in Global Mapper

A quick look at the LiDAR data confirms that it contains preexisting point classifications, including a lot of points marked as noise that look fine to me.

Raw LiDAR Data
Cleaning up and improving the classification of LiDAR points with the Automatic Classification tool.

My first task is to clean up and then improve the classification with the Automatic Classification tools. Using the Path Profile tool, which renders a lateral view of the point cloud data, I can clean the data up even more with some manual editing, since it is such a small area that I am interested in.

Classified LiDAR Data
Note the edges of the property boundary in blue on the profile window. There are some trees on both sides.

Applying Colors to a Point Cloud in Global Mapper

The Maine GIS site also provides 4-band ortho-imagery that was collected in a similar time frame to the publicly available LiDAR data. From that imagery, I apply the RGB color values to my point cloud using the Apply Color tool, which improves the point cloud analysis capability and creates an interesting visual perspective of the data. The imagery is leaf-off, so it does not match up perfectly with the point cloud, but it adds some detail that can help with identification and analysis.

View From the House in LiDAR Data
Looking down at the Kennebec River from my property with 3D colorized LiDAR points.
CIR
False color infrared (IR) display of the points highlights the coniferous vegetation and other late autumn greenery in red.
House Profile in False Color IR
Profile of the false color IR with the house in the middle.

Estimating Property Modifications with Global Mapper

After creating a terrain surface from the classified and filtered LiDAR data, I estimate the modifications that are needed to improve the drainage around the base of the house.

Using the new Breakline and Hydro-flattening tools, I create a flattened foundation by applying a height to the buildings in the terrain modeling process. Next, using the Watershed tool, I see the current drainage problem.

Drainage area that flows through the house and garage
Drainage area that flows through the house and garage shown in pink.

By using the digitizer tool and calculating the elevations, I create a line for a back drainage that would allow water to flow from start to finish. Then using buffering and site planning tools, I create a modified terrain surface that will calculate the necessary terrain modification.

I create a line for a back drainage that would allow water to flow off of the property.

Finally, I measure the volume of soil to be removed, and calculate the benching and terracing for the back retaining wall.

Site Plan Volume
Measuring the volume of soil to be removed for the drainage plan.

After the modification, the drainage from the back of the house to the road is much better. I am also glad to have some warning of just how much dirt removal a plan like this will involve.

Cross sectional path profile view of property
A cross sectional path profile view shows the new drainage line compared to the original terrain and classified LiDAR data.

 

Modified Drainage Watershed
Flow modeling shows how the terrain modification improves the flow of water around the back of the house.

 

I am still considering options for creating a small pond, ending with a tile drain, and many other possibilities. But thanks to freely available data and some quick calculating and visualization with Global Mapper, I have a much better sense of the scope of this project and what the final results might look like.


Katrina Schweikert
Katrina Schweikert


Katrina Schweikert is an Application Specialist at Blue Marble Geographics. She provides technical support, training, and software documentation. Katrina has over five years of professional experience in GIS, a GIS certificate from University of Wisconsin-Madison, and a degree in Geography from Middlebury College. She is happy to be working in technology back in her home state, as well as meeting GIS users across the globe.