With the development of unmanned aerial vehicle (UAV) technology

visible images are playing an important role in the maintenance of power systems. To achieve the shed breakage evaluation of composite insulators by UAV visible images

an intelligent fault assessment method is proposed. First

the composite insulators in visible light images are identified by Faster-RCNN. After image preprocessing

the image is enhanced and the noise is removed. Then

a canny operator is used to extract the edge of the sheds. An Improved Randomized Hough Transform (IRHT) is used to detect the ellipses in the edge image. The parameters of the detected ellipse

length of major axes and minor axes

center coordinates and deflection angle of major axes

are used to realize the segmentation of the composite insulator. Finally

the number of pixel points in the ellipse and the distance between the points and the ellipse boundary are used to judge whether there are breakage or cracks on the sheds. The area ratio of the breakage to the whole shed is calculated based on the number of pixel points inside the broken area. This method can be realized without a large amount of training dataset of the specific fault type and provides a technical basis for the online fault assessment of a composite insulator on overhead transmission lines.