Airffic Aircraft InstituteAircraft, evidence and the market
HomeFrom flight to workDoes a sharper photo make a more accurate map?

From flight to work / 02

Does a sharper photo make a more accurate map?

Image detail and positional accuracy are different, and accuracy needs independent checks.

5 minResearch reviewed through 5 November 20264 sources
On this pageA pixel size is not a position guaranteeA reconstruction needs reference informationCheck with points that were not used for fittingMake accuracy easier to demonstrateFollow the cause and effectSources for this lesson
Sir Airffic Go
Sir Airffic Go

A sharp-looking map can still put a road in the wrong place. Let us separate detail from accuracy.

Illustrative example

A pixel size is not a position guarantee

A digital photograph is made of pixels. Ground sampling distance, or GSD, describes how much ground one pixel represents. A smaller GSD can reveal finer detail, but it does not establish that each visible feature has the correct coordinates on the Earth. These are separate measurements. Coordinates express position numerically within a reference system; they serve a different purpose from a pixel count.

Imagine a detailed map shifted sideways as a whole. The lines remain sharp, yet every marked location is displaced. This is an illustrative example, not a reported experiment. Enlarging the map would make the lines easier to see without correcting their positions.

[workflow-p01]
From the evidence

A reconstruction needs reference information

Photogrammetry reconstructs shape by matching features visible in overlapping photographs. Known ground locations or aircraft positioning information help relate that reconstruction to real-world coordinates. The resulting map therefore depends on both the images and the reference information used to place them. Ground reference points supply a positional reference rather than simply adding image detail.

Blur, weak overlap and unsuitable viewing geometry can make reconstruction harder. A higher-resolution camera does not automatically remove these problems. The camera, flight pattern, reference information and processing form a measurement system, so evaluating one component alone leaves important sources of error untested.

[workflow-p04]
Detail needs an independent accuracy check
Images and references
Estimate shape and coordinates
Check independent points
Judge fitness for the task

Independent points were not used to fit the map; report the scope and conditions of the check.

From the evidence

Check with points that were not used for fitting

A study at a Spanish mine tested different combinations of surveyed control points. Errors measured only at points used to fit the model could look much better than errors at independent check points. The study supports separating model adjustment from the evidence used to evaluate its accuracy.

An analogy is judging learning only with exercises whose answers were already supplied. Independent checks test something different. They should also represent the area being mapped. The study involved one large, uneven site, so its numerical point-count thresholds should not become universal instructions for other surveys.

[workflow-p01]
Three different qualities
MeasureWhat it tells youWhat it cannot establish alone
GSDGround represented by a pixelCorrect coordinates
Position errorDifference at checked locationsComplete coverage
CoverageArea that was capturedDimensional accuracy everywhere
Our interpretation

Make accuracy easier to demonstrate

WingtraGROUND packages ground receivers and checkpoints with guided setup and data synchronization. Strategically, it targets the difficulty of establishing dependable survey evidence around the aircraft. Convenient guidance can reduce fragmented work, but users still need to examine whether the finished survey meets their task requirements.

LiDAR, which measures distances with laser light, also depends on orientation and sensor timing. Neither the word “LiDAR” nor a fine pixel specification settles the comparison. Define the required accuracy first, then compare demonstrated results, coverage and remaining uncertainty for the actual job.

[workflow-n04][workflow-n03][workflow-p04]
The idea to keep

Check visible detail, positional error and missing coverage separately.

Try the idea yourself

When two products say “one centimeter,” are they measuring the same thing?

Sources for this lesson