Generative Adversarial Networks (GANs) are one of the most popular Machine Learning algorithms developed inrecent times, and are a class of neural networks that are used in unsupervised machine learning. The advantageof unsupervised machine learning approaches such as GANs is that they do not need a large amount of labeleddata, which is costly and time consuming. GANs may be used in a variety of applications, including imagesynthesis, semantic image editing, style transfer, image super-resolution and classication. In this work, GANsare utilized to solve the single image super-resolution problem. This approach in literature is referred to assuper resolution GANs (SRGAN), and employs a perceptual loss function which consists of an adversarial lossand a content loss. The adversarial loss pushes the solution to the natural image manifold using a discriminatornetwork that is trained to dierentiate between the super-resolved images and the original photo-realistic images,and the content loss is motivated by the perceptual similarity and not the similarity in the pixel space. Thispaper presents implementation of SRGAN using Deep convolution network applied to both the aerial and satelliteimagery of the aircrafts. The results thus obtained are compared with traditional super resolution methods. Theresulting estimates of SRGAN are compared against the traditional methods using peak signal to noise ratio(PSNR) and structure similarity index metric (SSIM). The PSNR and SSIM of SRGAN estimates are similarto traditional method such as Bicubic interpolation but traditional methods are often lacking high-frequencydetails and are perceptually unsatisfying in the sense that they fail to match the delity expected at the higherresolution.
展开▼
