It may seem churlish to have a go at the competition, but we have become increasingly concerned at the quality and utility of products being offered by the second-tier and lower contractors appearing in the aerial photography market. In a market where pricing is a sensitive issue, some of these companies are delivering cheap imitations. Agencies who commission these services are finding out too late that the product is not as described on the packet. Remember: It is a long wait for the next budget cycle.
Rather than name names, we will focus instead on the issues that separate the sheep from the goats, so to speak.
For the spatial market, including GIS users, spatial accuracy of imagery is paramount. Why? Because more than any other GIS layer, imagery provides the best spatial framework for registration of other datasets. Spatial accuracy is essential for temporal studies—there is no point in trying to analyse global temporal change over misregistered datasets. Imagery is also often used as a base for digitising— again, no point in investing hundreds of hours work if the imagery is of poor spatial accuracy to begin with.
So, how to detect spatial accuracy errors?
Look at the repeatability of the system, multiple epochs over the same location are a good source for this.
Look for seamline errors.
Look for relief displacement errors in areas of high terrain variation.
Check against known survey points.
Check against cadastre of known accuracy. (Don‘t expect error-free cadastre — it often isn‘t.)
Spatial accuracy of imagery should be no worse than +/- 2 pixels.
Multi-ray matching and metric cameras
Multi-ray matching is the new buzzphrase in photogrammetry. The technical advantages gained by taking many images of the same point from different angles are being explored enthusiastically by the photogrammetric com-munity.
For aerial mapping applications, however, the benefits of multi-ray matching are only fully realised by using metric cameras. By this I mean fixed focal-length cameras that have been precisely calibrated in a laboratory and over aerial test ranges to determine the lens distortion characteristics and the overall metric stability of that specific camera. Metric cameras are not cheap. They are not off-the-shelf commercial-grade or pro-sumer cameras. Ask to see copies of the camera calibration certificate for each camera used in this survey.
Large-format vs Medium-format
There are some metric medium-format cameras on the market, but think carefully before you commission a substantial mapping project with medium-format. Large format is much more likely to enable the contractor to get the job flown and to deliver a quality result. The industry trend is for larger, not smaller, formats.
Does the system conform to proper standards of OH&S? Is the contractor flying with the aircraft door removed, with a camera that is suspended outside the aircraft fuselage, or with no cabin pressurisation over 13,000‘? Why would you commission a contractor to do something that is inherently unsafe?
Orthophoto mosaics should be completely free of visible joins and colour mismatches between frames. Anything less is not ortho standard, it is a quick-response Fastlook product and should be price compared accordingly.
Minor image blemishes such as excessive water flare, truncated trees, excessive building lean, wobbly bridges, etc, should all be dealt with prior to delivery of the final products. Bridges especially should be modelled at the top and the bottom of the crossover, and the two ortho images combined to form the final presentation.
Some of the more ludicrous ones of late:
“It’s super high resolution and you can zoom in and see a walnut on the ground with-out it pixelating.”
“Other contractors take several months to deliver imagery; only we can turn around imagery in such short time frames”
Wrong. Aerometrex can turn around its Fastlook imagery in days if needed.
“We alone can generate high-resolution imagery”
Spectacularly wrong. Aerometrex‘s 2.5cm GSD imagery is not only of extremely high resolution, it is also metric imagery, meaning it can be used for engineering applications.
“Other company‘s processing systems are highly manual”
Wrong. Most processing of aerial imagery is automated. The manual component comes in only when needed to apply final corrections. If you are not worried about final corrections or data quality, it‘s 100% automatic.
Beware of making a very courageous decision.