Overhead line forest fire identification method

The invention belongs to the field of intelligent operation and inspection of overhead lines, and particularly relates to an overhead line forest fire identification method, which analyzes color components of an image background by using an HSV color space, filters and extracts a suspected forest fire-containing picture by using a specific color threshold, effectively overcomes the defect that the color information of the picture is difficult to identify at night, and improves the identification efficiency. The identification rate of the forest fire at night is improved; performing region segmentation through a symmetric cross entropy fusion adaptive threshold segmentation technology, and removing interference of similar color components by using an expansion and corrosion morphological image processing method; the distribution difference of different regions is calculated through KL divergence to select connected domains, then adjacent contours are fused, the connected domains which do not meet the area thre