Detection of Micro-Cracks in Electroluminescence Images of Photovoltaic Modules

Abstract

This paper presents detection of micro-cracks in solar cells using Electroluminescence (EL) images. The preprocessing step in this work involved separation of solar panel section from background of EL image, use of perspective transformation, and separating individual solar cells from the Photovoltaic (PV) panel. Discrete Wavelet Transform (DWT) and Stationary Wavelet Transform (SWT) are used to extract textural features from these solar cells. These features were then used for classification of solar cells into cracked and non-cracked cells using Support Vector Machine (SVM) and Back Propagation Neural Network (BPNN). The networks were trained with a dataset of 2000 EL images and tested with a dataset of 300 test images. The percentage classification accuracy obtained is 92.67% and 93.67% using SVM and BPNN, respectively.