Selection of appropriate multispectral camera exposure settings and radiometric calibration methods for applications in phenotyping and precision agriculture

The radiometric accuracy of multispectral images is crucial for quantitative precision agriculture and phenotyping applications. Very often the effect of exposure time and gain on radiometric accuracy is overlooked, and most application acquire images using the camera's auto-exposure settings. However, the large scene-to-scene variations in exposure time and gain in response to the reflective range of objects on ground significantly affects the radiometic accuracy of the auto-exposure images. On the other hand, the use of fixed-exposure settings result in spatiotemporally consistent radiometric measurements from multispectral images. This dataset will provide the necessary information to replicate the results of our published work (https://doi.org/10.1002/ppj2.70000), which includes studying (1) the variations in exposure settings for different objects in the scene, (2) potential radiometric errors when using auto-exposure settings, (3) the ideal exposure range  for the different multispectral bands, (4) the spaiotemporal radiometric accuracy of auto- and fixed-exposure settings, and (3) the performance of auto- and fixed-exposure settings for cotton nitrogen monitoring. 

Methods
The images were collected with a Micasense RedEdge-3 multispectral camera mounted on a DJI Matrice-100 unmanned aerial vehicle. The script to process the data can be accessed from https://github.com/VaishSwami/Mutlispectral-camera-exposure-settings