By Greg Barnes
Recently, 2cofly had the great privilege of going to a few summer school programs for grade school students, and not only speaking about drones or doing a quick flight, but guiding students in running an actual drone mission. Our mission goal was to have the students create a small orthomosaic map of the campus. You can read a full write-up of everything we did, and see the orthomosaic maps the students created, here.
At the end, we compared our orthomosaic maps to Google Maps, and all the students agreed: Our map was clearer, sharper, brighter… just better overall than Google Maps.
So, what makes our map an orthomosaic map? First, I’ll need you to think about what a normal overhead perspective looks like.
Perspective view and other distortions
Let’s say you’re in a helicopter flying over a large city or forest, where there is an abundance of tall objects such as buildings or trees. Those tall objects, for the most part, rise straight up. So if you’re in a helicopter above these objects and you look directly below you, you’ll see that those trees or buildings appear to go straight down.
Again, the ones exactly directly below you will appear straight. But you’ll also notice that the objects not directly below you appear slanted — and they’ll appear more slanted the further they are off to the side.
These images illustrate this. Notice how in these normal top-down images, the tall objects near the center of the image appear straight, while the objects near the edges appear as though they are leaning — and the closer they are to the edge of the photo, the more they appear to lean.
This perspective view is present whether you’re looking at something with your own eyes or even using a photo.
But if you are looking at photos, other problems come into play. For instance, lens distortion and camera tilt. (I absolutely hate bad lens distortion. Recently I went through a friend’s online photo album on social media, and I was appalled by the terrible distortion on all of their photos. I need to find out what phone they have so I can avoid it at all costs.) *ALL* lenses have some amount of distortion or another, some more or less than others. Generally, higher-end lenses will have less distortion, and cheaper lenses will produce more distortion, but again, there is distortion in every camera due to the physics of light as it passes through the curved glass of the lens. There are different kinds of lens distortion. Here is a good image to show this.
The problems with Google Maps
Because of these various types of distortions, you can’t trust normal photos to be accurate if you’re trying to make precise measurements. For instance, if you are running a construction site, and you are relying on Google Maps for an overhead view and to make linear measurements (…or, really, if you are relying on Google Maps for much of anything), you will encounter problems due to a number of issues with the map you are using:
Google Maps is outdated, often by a matter of years. This doesn’t matter much in terms of residential neighborhoods or for a normal person getting directions, but with construction sites where things rapidly change, up-to-date imagery is essential.
Google Maps images are incredibly low quality. The public perspective may be that Google Maps are actually quite good in quality, but think about it: They’re made with satellite imagery taken from an altitude of 300 miles (woah). They really are poor quality. As a result, you are only able to zoom in so far, and even when you do, the image appears very blurry.
Google Maps images are off from where they should be, by a matter of feet — often entire car-lengths or more. This means that GPS coordinates you get from Google Earth will likely be off by as much. Again, not a big deal for the average consumer, but for industrial uses, this can be a big problem.
Google Maps images retain that distorted perspective view, meaning it will not be possible for you to take precise measurements. (Of course, Google Maps will suffice if you only need a general idea of distance. But people who need accuracy, such as those in the construction industry, would need something more.)
Orthomosaic maps are the solution
That's where orthomosaic maps come in: orthomosaic maps don’t have any of these problems. As our drone flies overhead, it takes many, many nadir images (that means, with the camera facing 90º down), and all these images have a very high level of overlap. This means that most of what you’re capturing will be in multiple photos. A parked car, for example, or the corner of a building, or any other object in view, might be in 10+ different photos.
We run those drone photos through software that finds matching points of overlap between images and stitches the images together. Then, the magic happens: This powerful, computationally-intensive software actually corrects for those different types of distortions — perspective view (so everything in the image appears exactly top-down), lens distortion, camera tilt, and even topographic relief, meaning that even hills and other varying terrain appear flat. All of these corrected images are known as “orthorectified” images.
With all these corrections, the final image is a large, accurate GeoTIFF file with lat/long location data in literally every pixel. This file is suitable for taking measurements — measurements with centimeter-level accuracy. If you opt to have orthomosaic maps made on a periodic basis, these highly-detailed images can even be used for record-keeping — a sort of visual log or visual journal that records and keeps track of the progress of an ongoing project.
Above, I wrote 4 numbered problems with Google Maps. Let me go through those again, one by one, to explain just how much better orthomosaic maps are.
Orthomosaic maps taken by a drone provide you with an up-to-date visual representation of the site. They are relatively quick to create, so could theoretically have a new one as often as you like. No need to wait years for imagery to be updated, as you would with an online map service.
Orthomosaic maps are incredibly high quality. As opposed to satellite imagery which, again, is taken from 300 miles, our drone shoots photos from less than 400 feet above. Typically, we would fly between 200-350 feet for these types of missions, depending on the amount of quality needed. And the quality we get is astounding. We recently uploaded two of those orthomosaic maps for the public to check out: One of Finegayan Elementary School, and the other of the University of Guam Agriculture Building — both created by grade school students under our direct supervision. See for yourself the amazing quality. (Use the Google Chrome browser, since other browsers may not work. When the map loads, under the “Layers” tab on the right, toggle the checkbox on/off to compare our orthomosaic map directly against the underlying Google Maps imagery.)
If you do toggle that checkbox, you’ll see that our orthomosaic map and Google Maps are not aligned in terms of location. This is because Google Maps is off. Our orthomosaic maps are location-accurate.
This is the key here for people who need to be able to rely on their photos for measurements: orthomosaic maps correct that perspective view distortion I mentioned before, they correct for other camera-based distortions like lens distortion and camera tilt, and they even correct for topographic relief. Due to all these corrections, your orthomosaic map is highly suitable for precise measurements.
At 2cofly, we have made lots of orthomosaic maps. They are useful, as we said, in the construction industry which is our focus, but they are also used all over the world in various industries: real estate, law enforcement (say, of a crime scene, or a visual log of busy city areas), agriculture (crop monitoring, etc.), disaster relief (for instance, post-flood analysis), and more. If you need an orthomosaic map, reach out to us: We are proficient at creating these and quickly deploying them to you and your team, so you can start improving your workflow with one right away!
blog written by