Is there any technical or accuracy issue in georeferenced data, ie can a UAV captured image provide accurate/similar result with a satellite image in spatial analysis?
Hi,
RTK can deliver few cm accuracy. Few (4) GCP can help to further improve this accuracy. However, to use RTK need to know the coordinates of your (master) receiver on the ground. The station on the ground must have known coordinates to estimate (in real-time) the GNSS bias in that area and send it to the second receiver (rover) on the drone. The rover receives the corrections from the master and can therefore improve the accuracy of its positioning.
Considering the specific problem of RTK on drones we need to consider that:
1) the synchronization between GNSS and camera must be assured. If there is a delay between shutter and position recording, we will have systematic shifts between the 2 observations.
2) We need to understand where the GNSS coordinates refer to on the drone: in other words, has someone computed the relative position between GNSS antenna and camera? Do the recorded coordinates refers to the antenna or the camera?
Most of these issues are explained and discussed in: Article Accuracy Analysis of Photogrammetric UAV Image Blocks: Influ...
Well, your everyday consumer UAV with a standard GPS on-board will of course feature an important localisation inaccuracy due to the limitations of the built-in GPS. But if you use a mapping professional UAV with an RTK-GPS system, then you can achieve cm-precision.
I'm interested in this question, I'll follow it.
It is a good question and a good answer of Dr Geert De Cubber
Can you give me more details about RTK-GPS system ?
Regards
Juliane Krenz,
Thank you for your response and the link you provided, it is very educating.
The earlier answers are correct and useful, but consider that also satellite imagery has limited geolocation accuracy - for example Worldview-3 with very high spatial resolution gives an error of some 3 m (unless additional precise ground control is used). If you then look at UAVs in the price segment of about 1000 Eur, such as DJI's Mavic or Phantom, you have a very comparable geolocation accuraccy. If you need better accuracy you can always incorporate some additional DGPS data, or indeed go for RTK. The latest Phantom has that built-in (and which gives you position accuracy of about 1cm [note: this is a claim by DJI, have not yet tested it myself, with another model, the Matrice 600 RTK we got rather poor accuracy of only about 50 cm).
Regards, Norman
See real examples in:
http://www.cropcopter.co/uav-imagery-vs-satellite/
Hi Norman,
I have been using consumer UAV to produce high-resolution orthophoto and DEM at volcanoes and I have compared georeferentiation without GCP to high-resolution digital-globe images and the error is typically about 2-3 m. As I don't intend to produce industry-level maps I am very happy about it. To my understanding RTK do improve the drone positioning but not necessary the product georeferentiation, you still need GCP to have
Hi,
RTK can deliver few cm accuracy. Few (4) GCP can help to further improve this accuracy. However, to use RTK need to know the coordinates of your (master) receiver on the ground. The station on the ground must have known coordinates to estimate (in real-time) the GNSS bias in that area and send it to the second receiver (rover) on the drone. The rover receives the corrections from the master and can therefore improve the accuracy of its positioning.
Considering the specific problem of RTK on drones we need to consider that:
1) the synchronization between GNSS and camera must be assured. If there is a delay between shutter and position recording, we will have systematic shifts between the 2 observations.
2) We need to understand where the GNSS coordinates refer to on the drone: in other words, has someone computed the relative position between GNSS antenna and camera? Do the recorded coordinates refers to the antenna or the camera?
Most of these issues are explained and discussed in: Article Accuracy Analysis of Photogrammetric UAV Image Blocks: Influ...
- Badges
- Science topic
- Similar topics
- Computer Science
- Images