• Uncategorized
• Posted By

Why Parrot ANAFI Ai is Made for Photogrammetry – Skyward

Much has been made about the Parrot ANAFI Ai being the first drone connected to Verizon’s 4G LTE network right out of the box. Being a connected drone gives the Parrot ANAFI Ai capabilities like redundant command and control in complex RF environments, enabling BVLOS operations, and uploading imagery for cloud processing while still in flight. 

Often overlooked amidst all of the discussion around its connectivity, however, is the Parrot ANAFI Ai’s inherent prowess at photogrammetry. In fact, Parrot says that their Parrot ANAFI Ai was “designed for large scale mapping.” So, what makes it such a photogrammetry machine? Aside from its autonomy and one-click flight planning, the Parrot ANAFI Ai’s 48MP camera is what makes it such a capable photogrammetry platform. It’s that camera that enables operators to capture the same or better image detail at higher altitudes, and fly faster while doing it that translates into a significant step forward in photogrammetry operations.

Parrot ANAFI Ai and the Importance of Ground Sample Distance (GSD)

Everything important about your 2D and 3D recreations — from how detailed they are to the accuracy of the measurements you can generate with them — is derived from the size of each pixel on the surfaces you’re viewing. This is called the Ground Sample Distance, or GSD, and is usually measured in centimeters per pixel (cm/pixel). Think of GSD as (a) how much area on the ground each pixel takes up, (b) the smallest detail it is theoretically possible to see within the end data product, and (c) the greatest measurement accuracy you can achieve. Because GSD is so important, let’s dig into what influences this measurement.

GSD begins with the pixels in the camera you’re using to capture your images, so let’s review some of the basics. The sensor in your camera is made up of an array of individual detectors we commonly call pixels. The total number of pixels on that detector is referred to as its resolution. So, if we look up the camera specs and see that the detector has 4,000 x 3,000 pixels, that’s a 12 megapixel (12MP) resolution detector. The higher the resolution of the detector, the more pixels you’ll have on target — assuming other factors like flight height, and camera focal length and field of view are the same.

When the camera is pointed straight down at the earth (called “nadir” orientation), each of those individual pixels covers a certain area on the ground from a given altitude. That is the GSD. If you position the drone higher above — therefore, further away — from the ground, each pixel will cover more ground. This larger GSD will result in less image detail and mean that you will not be able to see objects smaller than your GSD in your final data product. 

What impact does that have on you as a drone photogrammetrist?  First, you need to define how much detail is required for your end product before you fly. This could be determined by knowing the approximate size of the smallest objects you’ll need to be able to see clearly or measure accurately. But some inspections actually have technical standards you would need to comply with. 

Defining your GSD is important because that will be the factor which determines your maximum flight height above your target. To calculate either the height at which you’ll need to fly, or the GSD you’d achieve from a given altitude with a specific camera, Pix4D has a handy, free GSD calculator. 

Let’s say your desired GSD is 3cm/pixel based on the camera you’re using and the altitude at which you’re flying. Within these parameters each pixel will cover a 3cm-by-3cm area on the surface, the camera can only detect details 3cm or larger, and your measurements will only be accurate to within 3cm. (This is a rough number and can potentially be improved with compensations in the processing software.) The higher you fly from your target of interest, the greater area each pixel will cover, and the larger your GSD will become.  

Generally speaking, the smaller your GSD the better — you’ll have more image detail in your end product, and more accurate measurements. However, getting a smaller GSD value will probably require you to fly lower to your target area, and fly more legs of your flight pattern to cover the area adequately. Therefore, you’ll need to balance the GSD necessary for your mission against the time and effort you’re able to invest in capturing your datasets. 

Calculating GSD for Photogrammetry

Now that we understand GSD and its importance to photogrammetry, let’s look a little deeper into the math behind how it is determined. The calculation for GSD is:

GSD = (H×Sw) / (F×imW)

With:

GSD = ground sampling distance (in cm/pixel)

H = flight height (in meters)

Sw = sensor width (in centimeters)

F = lens focal length (in millimeters) [Note this is the camera’s actual focal length, not its 35mm equivalent.]

imW = image width (in pixels)

Realistically speaking, you’re not going to have to run these numbers yourself. You can always use Pix4D’s GSD calculator, or many flight apps have the GSD calculation built into them. With that in mind, let’s just look at how the different parts of the GSD calculation relate to each other. If everything else remains the same, but your flight height (H) goes up, your GSD will go up right along with it, because everything in the numerator of the equation has a direct impact on GSD — if either of them gets bigger or smaller, the GSD will follow suit. 

The elements in the denominator of the equation (focal length (F), and image width (imW)), on the other hand, have an inverse relationship with GSD. If either of them gets larger, GSD will get smaller. So, if everything else in the equation remains the same, but you go from using a camera with a detector that is 4,000 pixels across to one that is 8,000 pixels across, your GSD will get much smaller. And, as we’ve already discussed, a smaller GSD gives you more image detail from a given flight height.

And just like that, we’ve mathematically explained why the ANAFI Ai’s 48MP camera gives you so much better GSD — and therefore, photogrammetry — results than the 20MP cameras in common use on drones of this class.

Faster, More Detailed, Maps & Models

Let’s take this out of the theoretical and make it more practical. The Parrot ANAFI Ai’s 48MP sensor (8,000 x 6,000 pixels) lets you either maintain the same GSD you’d have with a lower resolution camera at a higher altitude, or fly at the same altitude with dramatically finer detail. Also keep in mind that flying at different altitudes will also impact how much ground the camera covers in a given image, called the camera footprint. The best of all possible worlds would be to maintain or improve upon your required GSD while flying with a larger camera footprint — this lets you get the detail you need, but fly much fewer passes over the target area and cover it in less time.

For example, let’s compare the Parrot ANAFI Ai’s GSD and resultant camera footprint with that of the Phantom 4 Pro to see how having a better GSD impacts the altitude at which you can fly, how much detail you capture, and how much ground you can cover on each leg of your flight pattern.

Example 1

	Parrot ANAFI Ai	Phantom 4 Pro v2
Resolution (pixels)	8,000 x 6,000	5,472 x 3,648
Altitude (feet AGL)	100’	100’
GSD (cm/px)	0.51	0.90
Image footprint	41 x 39	50 x 33

With both aircraft at 100’ above the ground, the Parrot ANAFI Ai (left) provides a better than 40% improvement in image detail.

Example 2

	Parrot ANAFI Ai	Phantom 4 Pro v2
Resolution (pixels)	8,000 x 6,000	5,472 x 3,648
Altitude (feet AGL)	100’	62’
GSD (cm/px)	0.51	0.52
Image footprint	41 x 39	29 x 19

While flying the Phantom 4 Pro v2 at an altitude that would give a roughly equivalent GSD to that of the Parrot ANAFI Ai at 100’ (left), the Phantom’s image footprint covers just over ⅓ the area as the Parrot ANAFI Ai, requiring you to fly more legs with the Phantom 4 Pro v2 to survey the same area.

Example 3

	Parrot ANAFI Ai	Phantom 4 Pro v2
Resolution (pixels)	8,000 x 6,000	5,472 x 3,648
Altitude (feet AGL)	100’	89’
GSD (cm/px)	0.51	0.74
Image footprint	41 x 39	41 x 27

Conversely, if the two aircraft were flown so that their cameras covered roughly the same area on the ground, the Parrot ANAFI Ai (left) would provide a nearly 50% improvement in GSD and image detail.

Conclusion

Making a decisive step forward by including a 48MP camera in a drone of its class, Parrot’s Parrot ANAFI Ai enables faster, more efficient, more accurate photogrammetry operations and results than similarly-sized drones with only 12 or 20MP cameras.

Reserve your Parrot ANAFI Ai now, and see how Skyward Mapping & Modeling, powered by Pix4D, can meet your drone photogrammetry needs.