Vision-Based Diagnosis and Location of Insulator Self-Explosion Defects

Abstract

Aiming at the insulator image of transmission line acquired by UAV, a vision-based insulator self-explosion defect detection and location method was proposed. First, superpixel segmentation is performed on the insulator image based on local texture features, and the saliency map of the insulator string is obtained by using the color feature different saliency and multi-scale optimization. Then, the salient image is binarized and morphologically processed to obtain a binary image. Finally, vertical projection Method to identify and identify the location of insulator defects. The experimental results show that the method can accurately identify the fault point of insulator strings. By comparing with two commonly used insulator self-explosion fault detection methods, the validity and reliability of the proposed method are proved.