Image Processing for daylight Electroluminescence PV Imaging acquired in movement

Regular photovoltaic (PV) system inspections have become a challenge with the significant growth in the number of modules and peak power capacity of PV installations. Image acquisition using drones, based on visual, thermographic, and more recently luminescence, can be a viable solution for large-scale PV inspections. As luminescence can provide a highly detailed and accurate PV module failure diagnosis, the development of a daylight electroluminescence (EL) imaging capability is of high importance. EL imaging performed in the field during the day requires the enhancement of the relatively weak luminescence signal over the noise from the sun. This is accomplished by image averaging and background subtraction, which requires the highly accurate registering of the of individual module images. A sequential EL image acquisition at high frame rates in continuous motion at different angles will be the realworld situation for a drone-based PV inspection in daylight, and to account for this movement while maintaining high quality images, several image processing steps must be developed. With this motivation, here we describe and perform EL image processing on a module with different faults to assure quality of the EL images in different motion speeds.