CNN-based Model for Cell Extraction from PV Modules with EL Images for PV Defects Detection

Electroluminescence technique has become a reliable technique for faults detects in the photovoltaic modules and it plays a key role in the process of module manufacturing and operation and management of photovoltaic plant. Given the inner structure of photovoltaic module, the defects of PV modules with EL images are all existed in the cell-level. Therefore, cell extraction is significant for the efficient and accurate detection of defects. In a PV module, a set of cells are connected and the borders between adjacent cells are not easy to detect due to the complexity of background in the EL image. In order to address the issue, a framework is proposed for the automatic cell extraction from PV modules in this letter. First, an image correction algorithm is introduced to correct the distortion; Then, the PV module is extracted from the complicated background based on the contour detection. Finally, a CNN-based model is proposed for cell segmentation and the proposed solution is evaluated from different aspects and the experimental results prove the effectiveness and superiority for cell extraction from photovoltaic modules.