Automatic edge identification for accurate analysis of thermographic images of solar panels

This paper proposes a segmentation method to analyze thermographic images of large solar fields with the aim to identify the borders of photovoltaic modules. This kind of analysis allows to correctly attribute possible faults, revealed by thermography, to a well specified solar panel. Moreover, the knowledge of the specific model of solar panel allows to bring the analysis to the single solar cell. This latter feature can help to eliminate some false positive that the mere observation of hot spots might produce.