ReflectDetect: A software tool for AprilTag-guided in-flight radiometric calibration for UAV-mounted 2D snapshot multi-camera imagery

摘要

Unmanned Aerial Vehicles (UAVs) equipped with optical sensors have transformed remote sensing in vegetation science by providing high-resolution, on-demand data, enhancing studies in forestry, agriculture, and environmental monitoring. However, accurate radiometric calibration of UAV imagery remains challenging. A common practice, using a single calibration target while holding the UAV-mounted camera close above it, is being criticized as the hemisphere is invisibly shaded and the reference images are not collected under flight conditions. ReflectDetect addresses these challenges by allowing in-flight radiometric calibration through automated detection via two different approaches: 1) a geotagging approach leveraging high-precision coordinates of the reflectance targets and 2) AprilTag based detection, a visual fiducial system frequently used in robotics. A brief statistical analysis and example data is provided to reassure the quality of the calibration results. ReflectDetect is available through a command-line interface and open-source ( https://github.com/reflectdetect/reflectdetect ). It now enables users to design new in-flight calibration studies to eventually improve radiometric calibration in applied UAV remote sensing.