Defect Detection of Catenary Suspension Insulator Based on Image Processing and YOLOX

The catenary suspension insulator is a critical part of the rigid catenary suspension device of urban rail transit. The defect detection of the insulator is of great significance to the normal power supply of urban rail transit. Therefore, it is important to automatically detect the defect of the insulators. In this paper, intelligent detection methods based on image processing technology and YOLOX are proposed. Firstly, the template matching algorithm is used to segment the insulators from the detected images, and then a model based on HOG features is proposed for insulator defect detection. Furthermore, to improve the detection accuracy and detection speed, YOLOX is trained for the defect detection and the identification of catenary suspension insulators. The experimental results show that the proposed methods can detect the defects of insulators with high accuracy and detection speed.