Aerial Surveys & Drone Mapping

Photogrammetry: how drones map the world.

Photogrammetry means measuring from photographs. The drone captures overlapping images, and software turns them into a model.

When a point appears in two or more photos from different positions, software triangulates its 3D position.

Out come an orthomosaic, a surface model and a 3D mesh. Their accuracy depends on four things.

GSD, ground sampling distance, is the size of one pixel on the ground. Fly higher and each photo covers more, but every pixel gets coarser.

GSD equals sensor width times height, over focal length times image width. Rearrange it to find your height.

For a Phantom 4 Pro at 2 centimetres per pixel, that's about 73 metres above ground.

Lower GSD gives more detail, but more images and longer processing. Match it to the job.

Quick check. You need finer detail. What happens to your flight height?

Fly lower. Finer GSD means a lower height.

Front overlap is 60 to 80 percent, 75 standard. Side overlap, between strips, is 60 to 70, 65 standard. Every point needs at least three images.

That sets the trigger interval. In the worked example, that's a shot every 27.4 metres: every 5.5 seconds at 5 metres per second.

Too fast for the camera? Slow down, or fly higher.

Buildings need an oblique orbit at about 45 degrees, plus a nadir roof pass. Corridors get a double grid, or one strip with very high overlap.

At the quarry, the crew lays ground control targets and surveys each one.

Ground control points have precisely surveyed coordinates. Without them, the shape is right, but position and scale can be off. With them, the model snaps into place.

Use at least five, around the perimeter, at high and low points, each in three or more images. Keep one or two as checkpoints.

The payoff: a stockpile model, with volume calculated from the surface model.

Recap. GSD sets your height. Overlap fills the gaps. The pattern fits the target. Ground control pins it down.

Final check. How many ground control points, and where?

At least five, spread around the perimeter.

Visual descriptions

The animation shows a shared point in overlapping photographs and triangulates its three-dimensional position. Outputs include an orthomosaic, a surface model and a mesh.

The GSD formula is sensor width times height above the ground, divided by focal length times image width in pixels. A Phantom 4 Pro worked example uses approximately 73 metres for 2 centimetres per pixel. GSD is image detail, not a guarantee of survey accuracy.

The first check asks how to obtain finer ground detail. The animation selects a lower flight height for the same camera.

The worked example illustrates 75 percent front overlap and 65 percent side overlap. It calculates a trigger spacing of 27.4 metres, or about 5.5 seconds at 5 metres per second. Actual overlap and capture settings depend on the site, sensor and processing requirements.

The quarry sequence shows surveyed control targets and a stockpile surface used for a volume calculation. Control points constrain the model, while separate checkpoints assess the results.

The final check compares three clustered points, at least five points around the perimeter, and one point at the highest location. The illustrated example selects at least five distributed points. Actual control and checkpoint numbers and locations depend on the project.