The Comparison of Solar Module Damage Texture Analysis using GLCM and LBP

Abstract

Solar modules play a vital role in renewable energy systems by converting sunlight into electrical energy. Over time, the surface of the panels can develop various issues such as cracks, scratches and stains, leading to a reduction in efficiency and energy output. Manual inspections have limitations in terms of time and cost; therefore, a solar panel damage detection system is required, utilising a reliable method for image analysis. The data used to test the model comprised 24 solar modules images, sourced from primary and secondary data. The collected images represent both physical and electrical damage. The methods used for feature extraction utilised the Gray Level Co-occurrence Matrix (GLCM) and Local Binary Pattern (LBP) techniques. GLCM features were calculated at four different angles (0°, 45°, 90°, and 135°), incorporating metrics such as contrast, dissimilarity, homogeneity, energy, and ASM, whilst LBP features were extracted using metrics such as mean, variance, and entropy. The process continued with damage segmentation of the images using Otsu Thresholding to calculate the proportion of damaged area. The results of the study show that the largest detected damaged area reached 35% for GLCM and 27% for LBP. These results indicate that GLCM is more effective in class separation, whilst LBP is capable of capturing local texture patterns. This model has the potential to support the automatic maintenance of solar panels and improve the efficiency of solar energy utilisation.