Drone Photogrammetry: From Photos to a Map

Drone photogrammetry: how a few photos become a map

In short

  • A single photo is just a picture; photos must overlap to become a three-dimensional map.
  • GCPs and SfM turn the photos into terrain with accurate coordinates.
  • Orthophotos, DSMs and point clouds are basic materials for decisions at a disaster site.

The day after heavy rain, a report comes in that part of a mountain slope has slid. The area is too dangerous to enter on foot, so a drone is sent up and brings back hundreds of photos. But opening these photos one by one on a monitor does not make a map. Behind the process that turns a pile of photos into accurate terrain information with coordinates are three mechanisms: overlap, reference points and computer vision. This post explains how drone photos become a "map" and how the results are actually used at a disaster site.

drone aerial mapping survey
Stock photo. Photo: Pexels / Pixabay

Why can't a single drone photo make a map?

A map is not a flat view from above; it is data with real coordinates and heights. A single photo cannot tell you exactly where a point is or how high it is. So the drone's flight path is planned to shoot the same point several times from several angles, and how much each photo overlaps its neighbor is called overlap. Usually the overlap along the flight direction (forward overlap) and across it (side overlap) are set separately. If the overlap is too small, the calculation described below fails altogether, or the terrain comes out distorted and bumpy. At sites with strong relief, such as a landslide, you should set more overlap than over flat ground so that shadowed areas and steep slopes are not missed.

Why do you have to place GCPs?

If you stitch the photos together, you get the shape of the terrain, but where that shape is on Earth is another matter. For that, you place markers on the ground at points whose coordinates are known exactly, called ground control points (GCPs), and record their precise coordinates with a GPS surveying instrument. You can make a rough map from the position information the drone stores itself. But to calculate damaged area, or real distance and slope for a report, you need results corrected with GCPs. Especially when estimating recovery construction costs or using the data as legal evidence, the margin of error often has to be stated, so whether GCPs were used and how many is a criterion for how reliable the product is.

How do photos become three-dimensional terrain?

The calculation method that turns hundreds of photos into three-dimensional terrain is called SfM (Structure from Motion). The principle is similar to how human eyes judge distance with two eyes. It finds the same point (a feature point) in overlapping photos, calculates how much that point shifts in position from photo to photo, and reconstructs the drone's positions and the heights of the terrain at the same time. The process runs in the order of feature extraction, matching between photos, camera position calculation and dense point cloud generation, and dedicated software usually handles it automatically. However, if there is fog or the photos are blurred, feature matching can fail and leave terrain with holes, so always check the processed result by eye.

How are orthophotos, DSMs and point clouds used at a disaster site?

When SfM processing is done, three products come out. The orthophoto is a photo corrected for distortion as if looked at straight down from above. The DSM (digital surface model) is height information that includes the ground surface plus buildings and trees. And the 3D point cloud represents each point as a dot. Orthophotos are used to draw the boundaries of flooded extent or wildfire-damaged zones by eye, and DSMs are used to calculate the volume of soil runoff or terrain change before and after a collapse. Point clouds are useful for examining cracks in buildings or the shape of collapsed structures in detail. If you overlay these three, you can get a sense of "where it collapsed and how much, and where the water flowed" without entering the site, so they are often used in initial response meetings.

What rules should you check before flying?

Drone shooting is not something you can do immediately just because you have the equipment. Depending on the region, flying itself may need prior approval, and shooting near certain facilities may be restricted or require a separate permit. Even at a disaster site, exceptions do not automatically apply, so you must check the rules of the responsible authority and the flight zone at the shooting-plan stage. Do not judge this from an internet search alone; it is safer to ask the responsible aviation authority or the local government department directly and confirm the rules in force that day. If you skip this check and fly, you may be unable to use the data however good the footage is.

Step by step on screen

The figure below is a diagram that summarizes the order of the steps.

Steps to follow

  1. Get sample data
    Download a public example photo set, such as the one in the OpenDroneMap official repository, as a whole folder
  2. Prepare the processing program
    Install WebODM, open it in a browser and click New Project on the Dashboard
  3. Upload the photos
    Choose the photo folder with the Select Images button and click Start Processing
  4. Check the output options
    In Processing Options, tick Orthophoto, DSM and Point Cloud
  5. Check the result
    When processing is finished, preview Orthophoto, DSM and Point Cloud from the tabs on the left
  6. Load it in QGIS
    Choose Layer > Add Layer > Add Raster Layer, select the orthophoto file and confirm (menu names can differ by version; if you cannot find it, look at the right-click menu in the Layers panel)

Menu names may differ slightly between program versions.

Prompts you can copy

Draft a checklist of regulatory and safety items to check before drone shooting at a disaster

You are an assistant who helps a drone shooting officer at a public agency.
We plan to shoot the damage site caused by [target disaster] in [area name] with a drone.
Write a checklist in table form that includes: confirming flight approval, confirming restricted shooting zones, confirming weather conditions, whether GCPs are placed, and confirming the pilot's qualifications.
For each item, also write the responsible agency or the way to check it.
Do not write legal provisions or figures you are not sure of; mark them "needs confirmation with the responsible authority".

Replace the bracketed parts for your situation. Check the facts in any answer before you use it.

Try it yourself

  • Make an orthophoto yourself in WebODM from public drone example photos.
  • Open an existing drone orthophoto from a national land information platform or a local government spatial information service and compare resolutions.
  • Search the website of the responsible aviation authority in your area for the drone flight approval procedure.

Common questions

Q. Do I always need GCPs for drone photogrammetry?
A. For a rough map, the drone's built-in GPS is enough. But for accurate area and distance calculations, or when the data is used as legal evidence, results corrected with GCPs are better in terms of reliability.

Q. How much overlap should I set?
A. It depends on terrain relief and the purpose of shooting, so rather than a single figure, it is safer to check the values recommended in the equipment and software manuals and to check by eye whether the actual processing result has any breaks in the terrain.

Q. In a disaster, can I fly without flight approval?
A. Even in disaster response, the flight zone and procedures often have to follow the rules of the responsible authority, so you need to check the approval procedure with the responsible department in advance.

What comes next

Each drone photo looks like an ordinary landscape picture, but after overlap, GCPs and SfM calculation it becomes a map that holds the coordinates and heights of a disaster site. A later post in this series looks at how to put AI on top of the orthophotos and point clouds made this way to find damaged areas automatically.

This is an English adaptation of a post in the Korean GeoAI basics series. The original is here, and more learning materials are on GeoAI for Disaster Safety.

This post was drafted with the help of AI tools and reviewed by the author. Tool features and data portals change over time, so check the official documentation when you practice.

Comments

Popular posts from this blog

What Is GeoAI, and How Does It Help Disaster Safety?

Getting Started with QGIS: What to Do First

Whitebox QGIS Practice 26 — Elevation Statistics of Two Basins: Means of 2,000.1 m and 1,924.3 m, a 76 m Gap