PTSRGAN: Power transmission lines single image super-resolution using a generative adversarial network
Shahrzad Falahatnejad, Azam Karami, Hossein Nezamabadi–pour
Abstract
In recent years, the use of UAV images to monitor power transmission lines has become a popular method. However, the low quality of these images can present challenges for accurate monitoring. To address this issue, this study proposes a GAN-based model that enhances the resolution of UAV images captured from power transmission lines. This model utilizes a generator with a novel structure and loss function, which enables it to produce high-quality images with detailed edges and textures. In addition, the discriminator uses a new Siamese-network based architecture, making it capable of better distinguishing between real and fake images. Experimental results show that the proposed method outperforms state-of-the-art super-resolution models regarding producing high-quality images with finer details and higher values of PSNR, SSIM, and HaarPSI in shorter training and inference time.