A Deblurring Method of Electric Power Inspection Images Based on GAN

Abstract

With large-scale application of drones in Electric power line inspection, the electrical images captured by their cameras can be used for target recognition and defect detection. However, due to the relative movement of the camera and the shooting target and the high-frequency vibration of the helicopter and UAV platform, Electric power images will be degraded. This is not conducive to the training of the model and subsequent recognition and detection. In this paper, the deblurring of electrical power images are achieved by generative adversarial networks. The article first investigates the current research status of image deblurring, and then studies the application of generative adversarial networks in power imagery, and finally trains the model on the electrical power image training set, and experimentally verifies that the method is blurred in the power inspection scene Effectiveness.