What is orthorectification?
In an aerial photograph, buildings often sit slightly off from where they really are. That is partly because a camera views a 3D world from an angle, which creates geometric distortions. Orthorectification corrects many of them.
It uses:
- The sensor or camera geometry
- The sensor's position and orientation
- Terrain elevation (see DEM, DSM and DTM)
- Ground control information, where applicable
The result is an image in which locations are represented more accurately on the Earth's surface.
Why terrain matters
The top of a mountain is much closer to the camera than the valley below it. Because of perspective, features at different elevations appear displaced. This is called relief displacement, and a terrain model can be used during orthorectification to correct it.
Orthorectification in drone mapping
A drone captures many overlapping photographs. Photogrammetry can turn them into:
- Orthomosaics
- DSMs
- Point clouds
- 3D models
The orthomosaic then works as a spatially accurate map layer.
What is an orthomosaic?
An orthomosaic is a single, geometrically corrected image made by combining many overlapping photographs. Instead of hundreds of individual drone photos, you get one continuous image of the whole survey area.
Why not use the individual photographs?
A drone survey can contain hundreds or thousands of photographs. Each covers only part of the area and may differ in camera angle, lighting and perspective distortion, and the photos overlap. Photogrammetry software aligns and combines them into one continuous product.
How an orthomosaic is created
The workflow, simplified:
- Drone flight
- Overlapping photographs
- Image alignment
- Tie points and photogrammetric reconstruction
- Point cloud
- Surface model
- Orthorectification
- Orthomosaic
What it is used for
Orthomosaics are widely used in agriculture, construction, mining, surveying, infrastructure inspection and land mapping. They are intended as spatially corrected, georeferenced mapping products, which is what makes them useful inside GIS.
What is ground sampling distance (GSD)?
Drone imagery is often described as having, say, a 2 cm GSD. GSD (Ground Sampling Distance) is the approximate size of one image pixel on the ground, so 2 cm GSD means each pixel represents roughly a 2 cm × 2 cm area.
Why GSD matters
GSD tells you how much spatial detail the imagery holds. A smaller GSD generally means finer detail: 2 cm GSD shows more than 10 cm GSD.
GSD vs accuracy
Overall accuracy also depends on:
- GNSS positioning
- Ground control points
- Camera calibration
- Flight conditions
- Image quality
- Processing quality
- Terrain
Popular tools
You do not need all of these to start.
| Name | What it is | Good for |
|---|---|---|
| Pix4D | Commercial photogrammetry software | Orthomosaics, DSMs and point clouds from drone photos |
| Agisoft Metashape | Commercial photogrammetry software | The same workflow, widely used for drone surveys |
| OpenDroneMap (WebODM) | Free, open-source drone image processing | The same outputs without a licence cost |
| QGIS | Free, open-source desktop GIS | Using the finished orthomosaic and DSM inside GIS |
Glossary
| Term | Meaning |
|---|---|
| Orthorectification | Correcting geometric distortions in imagery so locations are more accurate |
| Relief displacement | The apparent shift of features at different elevations caused by perspective |
| Photogrammetry | Measuring surfaces and objects from overlapping photographs |
| Tie points | Features matched across overlapping photos to align them |
| Point cloud | A set of 3D points reconstructed from the images |
| Orthomosaic | One corrected image built from many overlapping photographs |
| GSD | Ground Sampling Distance, the ground size of one image pixel |
| GNSS | Satellite positioning used to locate the drone and control points |
Test yourself
Try to answer each question before opening the answer.
1. What is relief displacement?
Features at different elevations appear displaced in an image because of perspective. A mountain top is closer to the camera than the valley below it.
2. Why is a terrain model used during orthorectification?
It lets the process correct for relief displacement.
3. Why build an orthomosaic instead of using the individual drone photos?
The photos each cover part of the area and vary in angle, lighting and perspective. An orthomosaic combines them into one continuous, corrected, georeferenced image.
4. Does a 2 cm GSD mean the map is accurate to 2 cm?
No. GSD describes sampling detail. Accuracy also depends on GNSS positioning, ground control points, camera calibration, flight conditions, image quality, processing and terrain.
5. What steps lead from a drone flight to an orthomosaic?
Overlapping photographs, image alignment, tie points and photogrammetric reconstruction, a point cloud, a surface model, then orthorectification.
6. What products can photogrammetry make from overlapping drone photos?
Orthomosaics, DSMs, point clouds and 3D models.
Quick quiz
Pick an answer for each question to see if you got it right.