The benefit of using computer vision for ecological, aerial remote sensing is that it is not necessary to have any information about the position and orientation of cameras in space. This is why it is possible to use relatively inexpensive and small, light-weight platforms for measuring vegetation canopy structure and possibly even diversity.
To make this process work, it is necessary to collect highly overlapping photographs over the forest canopy. It proved very difficult to do this in a systematic way when using kites and standard R.C. airplanes, but is possible with platforms like the Mikrokopter Hexakopter, shown at right.
Currently, image acquisitions with the Hexakopter cover 250m x 250m (820ft x 820ft) areas at maximum, at a maximum altitude of 100m (328ft) above the ground surface.
We developed this image acquisition strategy to make use of the Hexacopter's ability to fly a controlled route at a relatively constant speed by using GPS, a magnetic compass, and a pressure sensor for altitude control though a combination of manual and waypoint flying.
Our current image acquisition strategy was developed using basic principles of photogrammetry to achieve a given amount of overlap per image based on flight speed and altitude above the ground. The flight path shown at left was generated in a GIS and then uploaded to the Hexakopter.
Image acquisitions have the following characteristics
- Flight speed: ~5 m/s
- Altitude: ~40m above the canopy, max. 100m above ground
- Swath width: ~50m, actual can vary depending on wind speed
- Endlap: min. 60%, typically >90%
- Endlap is the overlap between 2 consecutive photos and is varied by speed and altitude
- Sidelap: ~30%
- Sidelap is the overlap between photos in adjacent tracks and is varied by swath width