Abstract: In the photovoltaic infrared thermal image, the brightness of the hot spot and part of the high-temperature working area is very close. When the traditional threshold segmentation technique is used to extract hot spots, the working area is often separated to form false hot spots. Combining the advantages of Otsu algorithm and Sauvola algorithm, a hybrid threshold segmentation method based on weighted gray image is proposed. The brightness of the working area is reduced by weighting the gray-scale image, so that the contrast between the hot spot and the working area is enhanced and the visibility of the hot spot is improved. By using Otsu algorithm and the Sauvola algorithm, the gray-level image is binarized, and the mixing threshold is calculated according to the difference degree of the binary image. The experiments results show that this method is suitable for detecting hot spots of photovoltaic panels in hot working area and can segment hot spot accurately and effectively.