Edge Detection Method for Infrared Image of Porcelain Insulators based on Modified Generative Adversarial Network

The infrared thermography technology has during recent years played an increasingly important role in periodic inspections on tens of millions of porcelain insulators on overhead transmission lines. Edge detection is among others an important prerequisite for the completion of object recognition, semantic segmentation and feature extraction, laying a solid foundation for the accuracy of comprehensive diagnosis on porcelain insulators. A new edge detection method based on modified Generative Adversarial Network (GAN) is proposed in this paper, in order to further improve the anti-noise performance and detection effect of common edge detection methods dedicated for infrared image of insulators. A comparative evaluation including PSNR, IoU and FPR indexes is carried out, and the results reveal that the new method outperforms other common edge detectors to a considerable degree.