Local Feature Descriptor Based on Directional Structure Map for Improving the Hotspot Detection in the Multispectral Aerial Image of a Large-Scale PV System

• Published in: Lecture Notes in Electrical Engineering
• Main Author: Tan L.V.; Jadin M.S.; Osman M.K.; Bakar M.S.; Islam M.I.; Al Mansur A.; Ul Haq M.A.
• Format: Conference paper
• Language: English
• Published: Springer Science and Business Media Deutschland GmbH 2024
• Online Access: https://www.scopus.com/inward/record.uri?eid=2-s2.0-85205114016&doi=10.1007%2f978-981-97-3851-9_6&partnerID=40&md5=b6bcd28c0789ce820d38c6b7664b2109

To achieve long-term reliability and maximize the power output of photovoltaic modules, early detection of potential faults is of paramount importance. Aerial thermal image inspection is a common method used to identify and locate hotspots in these modules. However, this approach can be negatively impacted by noise, leading to inaccuracies in hotspot detection due to thermal reflections from the surrounding environment. To overcome this challenge, the paper proposes a solution that involves combining visual and thermal images of the photovoltaic modules for multi-spectral image matching. This proposed approach introduces the use of absolute structure map (SMi) and directional structure map (DSMi). By employing a histogram of the oriented gradient based on SMi and DSMi, each interest point’s local region is effectively described. The SMi undergoes processing with the Gabor wavelet filter, while the SMi is treated with the average filter to construct histogram bins. Ultimately, the normalized feature vectors are integrated. The paper conducts experiments to evaluate the performance of the proposed structure map feature descriptor, yielding impressive Precision and recall values of up to 0.82 and 0.97, respectively. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.