
Scott B.
Real Estate, Mapping, Inspections
Photogrammetry is measurement taken from photographs. An aircraft flies a planned pattern so that every point on the ground or on a structure appears in several frames from several positions, and software then finds the same points across those frames, works out where each photograph was taken from, and reconstructs the shape of what was photographed. The output is a point cloud, a surface model and an orthomosaic, which is a photograph with the geometry of a map.
How much detail is available is decided before the aircraft leaves the ground. The ground sample distance, meaning the real world size of one pixel, follows from the flying height and the camera, so flying lower gives finer detail at the cost of more images to fly, more time and more processing. A specification is a trade between detail, area and time rather than a setting somebody forgot to turn up.
The other half of the specification is georeferencing, meaning how the model is tied to real coordinates. That is done either with surveyed markers placed on the ground and captured in the imagery, or with a positioning system on the aircraft that records where each frame was taken, or with both.
The flying is the short part of the job. The area, the level of detail and the coordinate system are agreed first, because between them they set the flying height, the pattern and the overlap. On site, any ground markers are placed and surveyed before the aircraft goes up. Processing normally takes longer than the flight did, and on a large site it can take considerably longer.
Practitioners differ, genuinely and publicly, on how much ground control a job needs. Aircraft carrying survey grade positioning can produce horizontal accuracy that stands comparison with a flight controlled from markers on the ground, and whether that removes the need for the markers is an open question in the published literature rather than a settled one: a paper in the International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences puts the question in its title and treats it as unresolved. What is not disputed is that accuracy has to be demonstrated on points that were not used to build the model, so the useful thing to ask is not which method was used but how the result was checked.
Deliverables are normally an orthomosaic, a point cloud, a surface or terrain model, and often a textured mesh, together with a statement of what accuracy was achieved and how it was tested. That last item is worth asking for by name, because it is the difference between a model that looks convincing and one that can be worked from.
An ordinary photograph shows. Photogrammetry measures. Because the same point appears in several overlapping frames taken from different positions, the geometry can be recovered, so the result carries scale and position and can be measured off. A single photograph carries neither.
This is the one question in the field that practitioners genuinely disagree on. Surveyed markers remain the traditional answer, and aircraft with survey grade positioning have made it a live argument rather than a settled one, with peer reviewed work published specifically to ask whether markers are still required. The answer for a given job depends on the tolerance the output has to meet, and on whether that tolerance can be demonstrated afterwards.
No. It reconstructs surfaces that appear in the photographs, so ground under a canopy is not in the result because it was never photographed. Water and glass are the other hard cases: transparent and poorly cooperating surfaces are a known difficulty for the stage that works out where each photograph was taken from, and reflections degrade the result. Laser scanning is a different instrument that behaves differently, and it is a separate instruction.
Accuracy is a specification rather than a property of the method. State the tolerance the output has to meet and it decides the flying height, the control and the processing, and it can then be checked against points held back for the purpose. A survey commissioned without a stated tolerance cannot be assessed against one afterwards.
Wind and rain, as with any flight, and low sun, because long shadows across a site are difficult to correct for afterwards. Movement is the constraint people do not expect: the method works by finding the same point in photographs taken moments apart, so vehicles, plant and anything else that moves during the flight does not reconstruct properly.
Browse the full drone pilot directory or explore drone services by location.