Direct georeferencing of unmanned aerial vehicle photography and radar imagery with a low-cost real-time kinematic GPS.
The conventional (indirect) georeferencing of remote sensing imagery requires the use of control points that link known positions in the imagery to known positions in map coordinates. The number of control points depends on the amount of distortion in the imagery, method of transformation and desired level of accuracy, but it is often large. Overall, the collection of ground control points is a cumbersome and time-consuming operation, and almost an unrealistic one when it comes to the georeferencing and mosaicking of a set of images acquired from a small unmanned aerial vehicle (UAV). Indeed, small UAVs are typically flown below 400 feet above ground level and can rapidly collect an enormous amount of high-resolution images when mapping an area (i.e. over 500 images for every square kilometer).
The number of necessary ground control points can be substantially reduced using a direct georeferencing, an approach that is obviously better suited for UAV mapping projects. Direct georeferencing requires a Global Navigation Satellite System (GNSS) and an Inertial Measurement Unit (IMU) to directly measure the position and orientation of the imaging sensors on a remote sensing system in order to georeference their data. These are integral components of a UAV navigation system and the information they collect can, incidentally, be used for direct image georeferencing.
At Athabasca University, we have assembled several small UAVs, ranging from multicopters and EPP-foam airplanes with limited flight time (<1 hour) and payload weight (up to 2 kg) to larger fixed-wing aircrafts with extended endurance (6+ hours) and payload capability (up to 10 kg). We are using these UAVs for tracking springtime snowmelt timing and spatial patterns, monitoring the impact of climate change on permafrost landscapes in Nordic communities, and soil moisture mapping.
The Summer 2015 research project will consist in equipping our UAVs with a low-cost (<$150), lightweight real-time kinematic (RTK) GPS system, the NavSpark-Raw, and assessing its performance in achieving highly accurate direct georeferencing of UAV photography and radar imagery. The Navspark-Raw is an-Arduino-compatible board with onboard GNSS that can send out carrier phase raw measurements to a host computer to be applied RTK corrections in order to generate positions with centimeter-level accuracy. On several occasions throughout the summer, our smallest UAVs will be deployed over an experimental acreage located approximately 150 km west of Edmonton (Alberta), to test the RTK GPS/INS system and verify the positioning accuracy with and without ground control points.
