Geometric Rectification of Fire Remote Sensing from Small Unmanned Aerial Systems (sUAS) for Scientific Applications
Abstract: In determining the radiative properties of wildfires and other forms of burning biomasses, accounting for the area of coverage in an image is crucial to making accurate calculations. Where sensors aboard orbital platforms, are often limited in spatial resolution, sensors onboard near-surface unmanned aerial systems can provide swift, high resolution observations of fires and other environmental phenomena. However, flight hardware performance constraints of small unmanned aerial systems (sUAS) can occasionally render ideal suites of instruments unavailable. Obtaining accurate data values missing due to the lack of adequate instrumentation, therefore, requires creative use of data from other instruments for image rectification. To that end the orientation and position of the observation platform relative to the target are essential. Acceleration vectors in a cartesian coordinate system from inertial measuring instruments can be used to find angles of inclination; these angles can then be used to find other relevant data, such as above ground level (AGL) altitude and distances from a target. After calculating platform orientation and position relative to the target, image rectification matrices can be produced for transformation, such as the rotation matrix via the “Aerospace Rotation Sequence”. Other notable components for correction include a distortion matrix specific to the intrinsic attributes of the instrument being used for observation. These matrices are then implemented in taking observations and applying transformations to recorded images to account for the image-ground area of targets at a per-pixel level as it pertains to the inclination off-nadir, as well as the position in space of the instrument. The developed data structure and methods in place allow for more efficient data processing to find these data values after flights while reducing margins of error in the corrected images, as well as an increased degree of freedom when correcting images. This internship has enabled the development of an algorithm that can be used to implement the aforementioned methods both retroactively and in future deployments of sUAS-based observations for fire characterization by the NASA Fire Energetics and Emissions Research (FEER) team and others at NASA.
Allan Teer*, Dr. Charles Ichoku, Luke Ellison
Judging Forms – Official judges only