1. Creating digital terrain models using a
mini-UAS: methods and applications
Dr David Gillieson
SkyView Solutions Pty Ltd
PO Box 158
Atherton, Q 4883
www.skyviewsolutions.com.au

2. What is an UAS?
• Unmanned aerial system
– Remotely controlled airframe with autopilot and payload (eg.
drone)
– Remotely piloted aircraft with ground control station
• Many types (micro-, small, medium, large; planes,
helicopters; electric, petrol motors)
• Initially developed by Defence aerospace industries, but
now well established in civilian operations and more
accessible

3. Types of UAV
Predator Heron
Bat Zephyr

4. Unmanned Airborne Systems (UAS)
Orthophoto mosaic from one flight (55ha) Swinglet UAS

5. 3D imaging from stereo UAS photography
• Uses bundle block adjustment approach
• Prepare geotagged images with significant stereo overlap
• Relate image matchpoints to camera photogrammetry
• Create stereo model with accurate depth measurements

6. Point cloud from UAS photography
PO Box 158
Atherton, Q 4883
www.skyviewsolutions.com.au

7. Triangular mesh surface (TIN)
PO Box 158
Atherton, Q 4883
www.skyviewsolutions.com.au

8. Creating DEMs from
UAS photography – example from FNQ
• Three flights conducted at 120m AGL
• Photos geotagged and telemetry data created including
heading, GPS altitude, pitch, roll and yaw data
• Data used on a local computer to generate a coarse
mesh DTM and 50cm pixel orthophoto mosaic for
evaluation
• Geotagged photos and telemetry data uploaded to a
cloud computer at the Swiss Technical University in
Lausanne for final processing
• Products generated included an orthophoto mosaic with
5cm pixels, an orthoDTM, ASCII point file, camera
parameters report and accuracy report

9. Bundle block adjustment
• The software searches for matching points between
image pairs by analyzing all uploaded images (~230)
• Those matching points are used in a bundle block
adjustment to reconstruct the exact position and
orientation of the airborne camera for every acquired
image
• Based on this reconstruction the matching points are
verified and their 3D coordinates are calculated from the
UAS telemetry data
• Those 3D points are interpolated to form a triangulated
irregular network (TIN)

10. • Connectivity graphic showing quality and number of image keypoint
matches. Total of 232 images used in analysis

11. Orthorectified image mosaic

12. Bundle block adjustment
Statistics
Feature
total number of keypoint observations 1,104,692
total number of 3D points 440,908
mean re-projection error 0.793865 pixels (= 4cm)

13. Hillshaded OrthoDTM

14. Accuracy assessment
• Thirty correlation points on grid established with
RTK GPS
• Marker plates visible in UAS orthophoto mosaic
• Extract height values from DTM – average value
from 1m buffer around points
• Linear regression between RTK and DTM
values
• Thanks to Brendan Twine of Twine Surveys,
Atherton

15. Comparison between elevations from the DTM
and surveyed elevations
550
y = 1.0093x - 13.039
R2 = 0.9833
545
540
535
elevation from RTK
530
525
520
515
510
520.00 525.00 530.00 535.00 540.00 545.00 550.00 555.00 560.00
elevation from DEM

16. One metre contours

17. Half metre contours

18. Water flow direction

19. Water flow accumulation

20. Conclusions
• Mini-UAS can be used to generate high quality
orthophoto mosaics and digital terrain models of
small areas up to 100ha
• These can be captured in overcast conditions as
long as wind speed <10knots
• DTMs correlate well with RTK GPS surveys, and
have comparable resolution to LIDAR
• DTMs can be used for hydrological modelling