Research on Hot Spot Detection of Photovoltaic Panels Based on Image Processing

Hot spots caused by photovoltaic (PV) panel faults significantly impact their power generation efficiency and safety. Current PV hot spot detection methods face challenges such as low detection rates for small targets and poor generalization. To address these issues, this paper proposes a PV panel hot spot detection method based on image processing. Aerial infrared images of PV panels often contain many bright areas that are not hot spots, causing interference. The proposed method first segments the PV regions using the SAM (Segment Anything Model), and then detects hot spots using a gray threshold method. Initially, infrared grayscale images of PV panels taken by drones are preprocessed to reduce noise interference. Next, the optimized SAM model is used to segment the PV panel images, extracting individual PV panel images. Finally, the grayscale distribution of each segmented area is analyzed, and an improved threshold algorithm is used to detect hot spots. Experimental results show that this method can efficiently and accurately identify hot spot areas on PV panels, demonstrating significant advantages in detection accuracy and generalization performance compared to other image detection methods.