Image Preprocessing for Outdoor Luminescence Inspection of Large Photovoltaic Parks

Electroluminescence (EL) measurements allow one to detect damages and/or defective parts in photovoltaic systems. In principle, it seems possible predict the complete current/voltage curve from such pictures even automatically. However, a precise analysis requires image corrections and calibrations, because vignetting lens distortion cause signal spatial distortions. Earlier works on crystalline silicon modules used cell gap joints (CGJ) as calibration pattern. Unfortunately, this procedure fails if detection of gaps is not accurate or contrast images low. Here, we enhance automated camera algorithm with reliable pattern analyze quantitatively quality process. Our method uses an iterative Hough transform line structures three key figures (KF) separate detected busbars gaps. This allows identification all gaps, noisy disconnected edges PV cells exist potential induced degradation leads low between active area background. our dataset, subset 30 EL (72 each) forming grid (5×11) lead consistent results. We apply process 997 single module evaluate results random 40 images. After correction perspective correction, residual deviation ideal target points previously CGJ after applied correction. For 2200 control evaluation images, achieve less than equal 3 pixels. 50% points, 1 pixel reached.