Georeferenced images: how smartphones, drones and cameras turn photos into GIS data

Smartphones, drones and many cameras turn photos into GIS data by recording where each one was taken. A satellite receiver fixes the device's position, and the latitude, longitude and often the altitude are written into the image file's metadata as a geotag. GIS software then reads that geotag and places the photo on a map.
Key takeaways
- A geotag records where the camera was, not where the subject is. So a photo of a pole 20 meters away is pinned where you stood.
- According to GPS.gov, GPS-enabled smartphones are typically accurate to within 4.9 meters (16 feet) under open sky, and less accurate near buildings, bridges and trees.
- Location is stored in a photo's EXIF metadata as degrees, minutes and seconds. Drones often add more, such as gimbal angles and RTK status, in a second format called XMP.
- A drone with RTK or PPK corrections can position each photo to within a few centimeters. That is why drones, not phones, are used for survey-grade maps.
- An image with no coordinates, such as a scanned plan, can still be georeferenced by matching points on it to the same points on a map.
What is georeferenced imagery?
Georeferenced imagery is imagery tied to known locations on the Earth's surface, so it can be placed on a map. A geotagged photo carries one coordinate, the camera's position. A fully georeferenced image, such as an orthomosaic or an aligned site plan, goes further: every pixel has a real-world coordinate.
We use smartphones for pretty much everything these days. But did you know they can be used for capturing ground-based GIS data, along with devices like drones and 360 degree cameras? It's all thanks to the location information stored within the captured imagery.
Georeferenced imagery refers to images that are tied to specific geographic locations on the Earth's surface. Photos from a phone, camera or drone come with embedded geographic coordinates (latitude and longitude), often obtained using GPS (Global Positioning System) technology. That lets each photo be placed on a map at the point where it was taken. This transforms raw images into valuable spatial data that can be used for various applications, from urban planning, facilities management, and environmental monitoring, to disaster response.
GPS is the US system, and it is one of several satellite navigation systems known together as GNSS (Global Navigation Satellite Systems). Most of this guide says GPS, because that is the word devices and file formats use.
Geotagged or georeferenced?
The two words are often used as if they mean the same thing, but they do not. According to the USGS, a georeferenced map or image has been tied to a known Earth coordinate system, so users can determine where every point on it is located. A geotag sets a lower bar. A photo taken from the ground shows a scene, not a map. So only the camera's position is known, not the position of each thing in the frame.
How do latitude and longitude coordinates work?
Two numbers, latitude and longitude, locate any point on Earth in degrees. Latitude runs from 0 at the Equator to 90 degrees north or south. Longitude runs from 0 at the Prime Meridian to 180 degrees east or west. Together they give every point on the globe its own address, which is what a geotag stores.
Latitude and longitude
Geographic coordinates, specifically latitude and longitude, are used to define locations on the Earth's surface. Latitude measures how far north or south a point is from the Equator, while longitude measures how far east or west a point is from the Prime Meridian.
Coordinates are usually written in one of two ways. Decimal degrees use one number for each, with a minus sign for south and west, so the Sydney Opera House is -33.8568, 151.2153. Degrees, minutes and seconds, on the other hand, split each degree into 60 minutes and each minute into 60 seconds. That second form is how a camera stores a position, as the next section shows.
Some interesting coordinates:
- Eiffel Tower, Paris, France - Latitude: 48.8584° N Longitude: 2.2945° E
- Great Pyramid of Giza, Egypt - Latitude: 29.9792° N Longitude: 31.1342° E
- Sydney Opera House, Sydney, Australia - Latitude: 33.8568° S Longitude: 151.2153° E
- Niagara Falls, Canada/USA Border - Latitude: 43.0799° N Longitude: 79.0747° W
- Machu Picchu, Peru - Latitude: 13.1631° S Longitude: 72.5450° W
Small differences in the decimals matter. According to the USGS, one degree of latitude is about 69 miles. So the fourth decimal place (0.0001 degrees) is about 36 feet, or 11 meters, on the ground. For instance, the pair 43.0962° N, 79.0377° W, which this list gave for Niagara Falls until November 2026, lands in the city of Niagara Falls, New York. That is about 2 miles east of the falls themselves.
Coordinate systems and projections
To accurately place georeferenced imagery on a map, we use coordinate systems and map projections. Coordinate systems provide a framework for defining locations, while map projections translate the Earth's curved surface onto a flat map. No projection does that perfectly, however. According to the QGIS documentation, every map shows distortions of angular conformity, distance and area. Mapping software therefore uses a projection that keeps the distortion small for the area and the task.
A phone or drone records its position in WGS 84, the datum GPS itself uses. A site survey, by contrast, is often in a local projected system, such as a UTM zone or a state plane zone, measured in meters or feet. Our guide to coordinate reference systems explains how software converts between the two.
How does a smartphone, camera or drone geotag a photo?
A device geotags a photo by asking its receiver for a position at the moment the shutter fires, then writing that position into the image file. The latitude, longitude, altitude and time are stored as metadata, usually in a standard block called EXIF, so any GIS or photo software can read them later.
Geotagging
Geotagging is the process of embedding geographic coordinates into an image at the time of capture. This is commonly done using devices with GPS capabilities, such as smartphones and cameras. When you take a photo with your smartphone, it can automatically record the location, creating a geotagged image.
Location has to be switched on for the camera app first, though. In addition, a phone may fill in its position from Wi-Fi or mobile networks when satellite signals are weak. The EXIF standard even has a field for this, which records whether a position came from GPS, a cell ID, Wi-Fi or manual entry.
What does EXIF store about location?
EXIF (Exchangeable image file format) is the camera industry's metadata standard, published by the Camera and Imaging Products Association (CIPA) in Japan. Its latest edition, Exif 3.1, was published in January 2026. Location sits in its own GPS section of the file, and the main fields are:
- Latitude and longitude: Each stored as degrees, minutes and seconds, with a separate N or S and E or W reference.
- Altitude: Height in meters, with a reference that says whether it is above or below sea level.
- Time and date: The time of the position fix, in UTC.
- Image direction: The compass direction the camera was facing, where the device records it.
- Map datum: The datum the coordinates are based on, usually WGS 84.
- Positioning error: An estimate of the horizontal error, which some phones fill in.
The standard also separates the camera from the subject. The main latitude and longitude fields hold where the camera was, while a separate set of destination fields can hold the position of the thing photographed.
What is XMP, and why do drones use it?
XMP (Extensible Metadata Platform) is a second, more flexible metadata format. Adobe created it, and it has been an ISO standard (ISO 16684-1) since 2012. Unlike EXIF's fixed list of fields, XMP lets a manufacturer add its own. Drone makers use it to record what EXIF has no field for. For instance, according to the ExifTool documentation, DJI drone photos can carry absolute and relative altitude, gimbal and flight angles, and RTK status fields.
Cameras without a GPS receiver
Not every camera has a receiver. Some borrow the position from a paired phone. Alternatively, any photo can be geotagged afterwards from a GPS track log. Software matches each photo's capture time to the log and writes in the position for that moment. According to the ExifTool documentation, it estimates positions between log points by linear interpolation, so the camera's clock has to agree with the log.
If you're capturing data in the field, always make sure your equipment is recording location information. Then check that your software is reading that metadata correctly when you upload, because some apps and export settings remove it.
How accurate is the location in a smartphone, camera or drone photo?
It depends on the receiver and the sky above it. A smartphone is typically accurate to within about 5 meters under open sky, and worse near buildings and trees. A drone's standard receiver is accurate to the meter level, while one with RTK or PPK corrections can reach a few centimeters.
| Device | Where the position comes from | Typical horizontal accuracy |
|---|---|---|
| Smartphone | Its GNSS chip, helped by Wi-Fi and mobile networks | Within 4.9 m (16 ft) under open sky; 7 to 13 m in one urban study |
| Camera or 360 camera | Its own receiver, a paired phone, or a track log matched by time | Similar to a phone, or the phone's own accuracy when it borrows the position |
| Drone, standard GNSS | Its onboard receiver | Meter level |
| Drone with RTK or PPK | Its onboard receiver, plus corrections from a base station or network | A few centimeters |
| Survey-grade GNSS receiver | A dual-frequency receiver with corrections | A few centimeters in real time; millimeters from long-term measurement |
Smartphones
According to GPS.gov, GPS-enabled smartphones are typically accurate to within a 4.9 meter (16 foot) radius under open sky, and their accuracy worsens near buildings, bridges and trees. Real sites are rarely open sky. In a 2019 study on the University of Georgia campus, published in PLOS ONE, an iPhone 6 averaged 7 to 13 meters of horizontal error. Newer phones may do better, but the gap between open sky and a built-up site is the point.
A phone's own accuracy figure needs careful reading too. According to the Android developer documentation, the accuracy Android reports is a radius with 68% confidence. In other words, there is roughly a one in three chance the true position lies outside it. It also covers horizontal position only. As a result, it says nothing about the height in the same photo.
Drones
Drones are where the accuracy changes category. A drone with RTK (real-time kinematic) or PPK (post-processed kinematic) positioning uses corrections from a base station to tag each photo to within a few centimeters. GPS.gov describes the same jump: dual-frequency receivers with augmentation can position in real time within a few centimeters. Our guide to RTK and PPK geotagging explains how the two differ.
How accurate a photo's geotag is, however, is not the same as how accurate the finished map is. Processing, ground control and the flight itself all play a part, and our breakdown of drone data accuracy covers the whole chain.
The camera, not the subject
Remember, too, that a geotag places the camera, not the subject. A photo of a damaged pole taken from 20 meters away is pinned 20 meters from the pole, however good the receiver is. The image direction field, where the device records it, at least shows which way the camera was facing.
What metadata does georeferencing rely on?
Georeferencing relies on metadata, the behind-the-scenes information that tells software how to interpret spatial data. For a photo, that means its position, altitude, time and camera angle. For a finished map layer, it means the coordinate system, projection, height reference and pixel size. Without them, software cannot place the data correctly.
If you've ever opened a drone image and seen information about the GPS location, altitude or camera settings, that's metadata at work. Some key types of geospatial metadata that support georeferencing include:
- Coordinate system: For example WGS 84, NAD83 in North America or GDA2020 in Australia.
- Projection: For example a UTM zone, or an MGA zone in Australia.
- Altitude type: Height above the ellipsoid, or above the geoid, which sits close to mean sea level. The two give different numbers for the same point, so mixing them moves a whole survey up or down.
- Capture date and time: Which survey a layer belongs to, and the order of repeat visits.
- Camera position and angle: Where each photo was taken from and which way it pointed.
- Pixel resolution: The ground size of each pixel, known for drone imagery as the ground sampling distance (GSD).
Photogrammetry software reads the camera positions in drone photo metadata and uses them to place its outputs. That means drone photos captured with a GPS-enabled camera can be placed on the map with no manual work, though only as accurately as their geotags allow.
How are images without coordinates georeferenced?
Images without coordinates, such as scanned plans, old maps or drawing images, are georeferenced by matching points. You pick features you can see on both the image and a map that is already georeferenced, such as road intersections or building corners. Software then uses those matched control points to shift, rotate and scale the image into place.
Manual georeferencing
For images without embedded coordinates, manual georeferencing can be done using control points and reference maps. This involves identifying known locations within the image and matching them to corresponding points on a reference map to assign accurate geographic coordinates.
The number of points depends on how much the image needs to bend. According to the QGIS documentation, a simple shift and scale, or one that adds rotation, needs at least 2 control points. A general affine transformation needs at least 3, while second and third order polynomials need at least 6 and 10. More points are not automatically better, however. Esri's documentation advises spreading them over the whole image, ideally with at least one near each corner, rather than bunching them in one area.
Birdi (birdi.io), for example, works this way for drawings and images. Drone photos are placed by camera position, while a JPG or TIFF without coordinates is aligned by matching points on the Plus, Growth and Ultimate plans. Two points place it, three make it more accurate, and four or more show how far off each point is and flag any that do not fit.
Ground control and direct georeferencing
Drone surveys take the same idea further. Instead of matching points by eye, surveyors place ground control points with precisely measured coordinates, and processing software ties the imagery to them.
Alternatively, an RTK or PPK drone can georeference its imagery directly from its own recorded positions. Our guide to georeferencing for site surveys covers when each method is enough and how to check the result.
Can you map a site with smartphone photos?
Smartphone photos make good GIS point data, but they are a poor substitute for a drone survey. A geotagged phone photo puts an inspection, defect or asset on the map at the spot where it was taken. An orthomosaic or elevation model, on the other hand, is built from many overlapping images captured from above.
Ground-based photos, such as those captured by smartphones, are not the right input for orthophotos, DEMs or contours. For those, you'd need top-down imagery captured from above by drones. Our beginner's guide to drone mapping explains how a mapping flight captures that imagery.
Where phone photos earn their place is in recording what is at a location. Think of a crack in a retaining wall, a damaged fence panel, a pit lid or a blocked drain. Each photo becomes a point on the map with a time stamp, so it is easy to find again and to compare with the next visit.
If the photos need to line up with a drone map, check the size of the error against the job. For example, an error of 5 meters barely matters when you are finding a defect along a fence line. However, it can put a pit lid on the wrong side of a road.
Which device should you use to capture georeferenced images?
Choose the device by the accuracy the job needs. A smartphone is enough to log inspections, defects and site photos to within a few meters. A drone with standard positioning suits visual site maps, while volumes, earthworks and comparisons between surveys need RTK or PPK positioning, ground control, or both.
| Job | Device | Why |
|---|---|---|
| Logging defects, inspections and site photos | Smartphone or camera with location switched on | A few meters is enough to find the spot again |
| A visual site map or progress overview | Drone with standard GNSS | Overlapping top-down images, positioned to the meter level |
| Volumes, earthworks, as-built checks and repeat surveys | Drone with RTK or PPK, plus ground control | Centimeter positions, with an independent check |
| Putting an old plan or drawing on a map | A clean scan or export, then control points | The control points supply the coordinates |
Devices equipped with GPS functionality turn regular images into GIS data. The harder part, for many teams, is getting those photos, drone maps and drawings in front of the people who act on them. For that, Birdi is a sensible option for mid-sized construction and mining teams whose project managers, engineers and clients need the site data but do not fly or process it. A GIS analyst who needs a full desktop GIS will still want ArcGIS Pro or QGIS, however. Likewise, a surveyor who wants hands-on control of photogrammetry processing may prefer a dedicated processing tool.
Frequently asked questions
How do I find the location of a picture?
Open the photo's details or metadata. On most phones, the photo's info panel shows a map if the image has a geotag. On a computer, any EXIF viewer shows the GPS latitude and longitude. If no location appears, the camera did not record one, or it was removed when the file was shared or exported.
How accurate is smartphone GPS for GIS data collection?
According to GPS.gov, GPS-enabled smartphones are typically accurate to within 4.9 meters (16 feet) under open sky, and less accurate near buildings, bridges and trees. A 2019 urban study measured 7 to 13 meters. That is enough to log inspections and defects, but survey work needs an RTK or PPK receiver.
What is the difference between a geotagged and a georeferenced image?
A geotagged image carries one coordinate, the position of the camera when the photo was taken, so it appears on a map as a point. A georeferenced image is tied to a coordinate system so that every pixel has a real-world location, as in an orthomosaic. Therefore, only a georeferenced image can be measured like a map.
What is an example of GPS coordinates?
The Sydney Opera House sits at 33.8568° S, 151.2153° E. In decimal degrees, with a minus sign for south, that is -33.8568, 151.2153. GPS gives positions in the WGS 84 datum, and a camera stores the same position in its EXIF metadata as degrees, minutes and seconds, with an S and an E reference.
What is the difference between georeferencing and digitizing?
Georeferencing aligns an image or map to a known coordinate system so it sits in the right place. Digitizing, by contrast, converts the features on a map or image into digital coordinates, usually by tracing points, lines and polygons. You typically georeference a scanned plan first, then digitize the features you need from it.
Related reading
- Photogrammetry explained: how overlapping drone photos and their geotags become maps and 3D models.
- What an orthomosaic is: the fully georeferenced image a drone survey produces.
- Drone flight planning tips: how to plan a flight so its photos carry the overlap and positions mapping needs.
Sources
- GPS.gov. "GPS Accuracy." U.S. government, read November 2026. https://www.gps.gov/gps-accuracy
- Merry, K. and Bettinger, P. "Smartphone GPS accuracy study in an urban environment." PLOS ONE, July 2019. https://doi.org/10.1371/journal.pone.0219890
- U.S. Geological Survey. "What does 'georeferenced' mean?" Read November 2026. https://www.usgs.gov/faqs/what-does-georeferenced-mean
- U.S. Geological Survey. "How much distance does a degree, minute, and second cover on your maps?" https://www.usgs.gov/faqs/how-much-distance-does-a-degree-minute-and-second-cover-your-maps
- CIPA. "Exif Version 3.1" (DC-008-Translation-2026), January 2026. https://www.cipa.jp/e/std/std-sec.html
- ExifTool. "GPS Tags", "DJI Tags" and "Geotagging with ExifTool." Read November 2026. https://exiftool.org/TagNames/GPS.html, https://exiftool.org/TagNames/DJI.html, https://exiftool.org/geotag.html
- Adobe. "XMP documentation." Read November 2026. https://developer.adobe.com/xmp/docs/
- Android Developers. "Location." Read November 2026. https://developer.android.com/reference/android/location/Location
- QGIS Project. "Georeferencer." QGIS User Guide. https://docs.qgis.org/latest/en/docs/user_manual/managing_data_source/georeferencer.html
- QGIS Project. "Coordinate Reference Systems." A Gentle Introduction to GIS. https://docs.qgis.org/latest/en/docs/gentle_gis_introduction/coordinate_reference_systems.html
- Esri. "Overview of georeferencing." ArcGIS Pro documentation. https://pro.arcgis.com/en/pro-app/latest/help/data/imagery/overview-of-georeferencing.htm
- Esri. "Georeferencing" and "Digitizing." GIS Dictionary. https://support.esri.com/en-us/gis-dictionary/georeferencing
- Birdi. "Drawings and images as map overlays." Help centre, read November 2026. https://help.birdi.io/en/articles/17751040
