Drone Photogrammetry: How Drone Photos Turn into 3D Surveys

| September 17, 2020

Quick answer: What is photogrammetry?

Photogrammetry is the science of collecting physical information from 2D photos, often aerial images captured by drones.

By combining enough overlapping images of the same features, photogrammetry software can be used to generate photorealistic 3D representations of topographic surfaces.

Surveying and GIS professionals use drone photogrammetry, along with accurate GPS data, to create 3D maps and models for use in construction, waste management, mining, and aggregates workflows.

Producing accurate, useful 3D drone surveys requires three ingredients: aerial drone imagery, good ground control, and the science of photogrammetry. But the actual process we use to transform your 2D drone images into 3D site models for analysis in the Propeller Platform can seem mysterious. Even some of our customers who’ve flown drones on their sites for years aren’t clear on the finer points of photogrammetry.

So, we wanted to demystify the science that helps us deliver drone data success for earthwork operations around the world. (We’re going to focus on photogrammetry in this post. Check out our blogs on ground control basics and the secrets to capturing quality data to get an even better understanding behind drone surveying workflows.)

What is drone photogrammetry?

At its most basic, “photogrammetry” is measuring real distances via photos. But not just photos—A LOT of photos. More photos than you probably think is necessary. Think your parents taking pictures of you and your date before leaving for prom. You need to every angle possible.  

drone flying above construction site

Or, at least, you need enough photos taken from the optimal height to get you the best ground sample distance. In fact, you need an 80% overlap on each picture. This is necessary for two reasons:

  1. For the computer to stitch images together to make the orthophoto (2D aerial image that’s been corrected for distortion).
  2. To capture enough angles to create a digital terrain model (the shape of your surface).

When combined, the orthophoto and DTM create the 3D model of your site. 

We can’t overstate the importance of steady, consistent flight in getting these photos right for drone photogrammetry. (The best way to achieve that is with a flight planning app. We recommend DJI’s Ground Station Pro, which you can read more about in our Surveyor’s Guide to Drones blog series.)

Ultimate Guide to PPK Drone Surveying

The gist of the science: How does drone photogrammetry work?

If you see the same feature from three or more known positions, you can triangulate its location in space. In other words, you can capture its horizontal (x,y) and vertical (z) coordinates. A feature is any visually distinct point in an image.

point cloud of a substation

If you took an average image from your survey, you’d easily be able to pick out many “features” between images. The more features you match, the better you can relate images to each other and reconstruct objects within them. This is exactly what photogrammetry software does for one feature, and the next, and the next, and so on, until it’s covered your entire site.

Once you have a lot of these features—think millions—you can create a “cloud” of points. Each point has a matched feature describing your surveyed area in that location. You can then turn your point cloud into any regular outputs used in geospatial software, like a 3D mesh.

You’re using photogrammetry right now!

The best way to visualize this is to use your eyes—literally. Your eyeballs are using photogrammetry all the time.

You have two eyes (two cameras), processing a live feed of your surroundings. Because your eyes are slightly apart, you’re getting two different inputs at slightly different angles. (Test this yourself by holding up a finger in front of your face. Look at it with one eye closed, then the other. You’ll notice your finger jumps relative to background objects.)

cut and fill measurement using drone photogrammetry

Your brain knows how far apart your eyes are. This allows it to process this info into a sense of distance by merging both feeds into a single perspective. (If you’ve ever tried to catch a ball with one eye closed, you know the lack of depth perception makes it difficult.)

Your mind is rendering a live depth map of the 3D world from two 2D inputs—just like how Propeller renders a 3D survey from many 2D photos.

Drone photogrammetry used in surveying workflows—such as Propeller PPK—also processes geospatial information collected from the drone and ground control points. Doing so allows you to use your 3D survey to collect highly accurate real-world quantities—within 1/10 ft when done properly—with just a few clicks of your mouse.

Harnessing the Power of Propeller Hardware: AeroPoints, DirtMate, and More

How does photogrammetry software work?

Once you know the shape of something in 3D (from the features and points above), you match that up with its visual appearance. This is where image stitching, or “ortho-rectification,” come in. Stitching is like putting together a jigsaw puzzle. When you assemble one, you look for common features—corners, bright shapes, etc.—to fit together everything more easily.

Now imagine trying to do that if the puzzle was facedown or all shades of brown. Much more difficult, but that’s the level of complexity that comes with stitching photos of physical terrain.  Fortunately, computers are great at recognizing patterns.

road grade measurement with drone software

And when we feed them multiple images of the same feature from different angles, computers can compile the finished picture—the orthophoto—easier and faster than a human ever could.

But there are limitations. If a surface is too featureless or turbulent—like a building’s polished windows or the ocean—stitching doesn’t work very well. It’s impossible to match a feature between images if it’s there in one photo but not in the next, or if there aren’t enough hard edges or identifiable features to tell images apart.

Photogrammetry is also less effective when capturing images of sites with more vegetation. It’s one of the few area where lidar has the edge over photogrammetry.

Find out why drone photogrammetry is a smarter investment than lidar for most earthwork management.   

The final picture: What do you get from a photogrammetric survey?

Once those steps are completed, you’ve got two things: a point cloud and an orthomosaic. The former is where all your “terrain” outputs—DXF mesh, geoTIFFs, etc.—come from. The orthomosaic is then layered over the elevation model, which gives you that final measurable, 3D model of your worksite.

Once you have those models, what you do with them is up to you. You can use industry-specific measurement tools to analyze your site or export the data for use in other GIS or CAD software. However you plan to use your drone data, photogrammetry will get you the information you need faster, and with better accuracy, than traditional surveying methods alone.

Ready to leverage the full power of drone photogrammetry to survey your site in a fraction of the time? Let’s talk.

Construction Trends Report 2020

Subscribe to The Dirt e-newsletter

Get the latest dirt from Propeller, including industry news, customer stories, webinars, videos, and more.