Unsupervised Learning for Real-World Super-Resolution
This addresses the challenge of real-world super-resolution for applications lacking paired data, though it is incremental as it builds on existing unsupervised techniques.
The paper tackles the problem of super-resolution without paired low and high resolution images by proposing an unsupervised method to invert bicubic downsampling, enabling training with pixel-wise supervision and robust generalization to real-world images, as demonstrated in experiments.
Most current super-resolution methods rely on low and high resolution image pairs to train a network in a fully supervised manner. However, such image pairs are not available in real-world applications. Instead of directly addressing this problem, most works employ the popular bicubic downsampling strategy to artificially generate a corresponding low resolution image. Unfortunately, this strategy introduces significant artifacts, removing natural sensor noise and other real-world characteristics. Super-resolution networks trained on such bicubic images therefore struggle to generalize to natural images. In this work, we propose an unsupervised approach for image super-resolution. Given only unpaired data, we learn to invert the effects of bicubic downsampling in order to restore the natural image characteristics present in the data. This allows us to generate realistic image pairs, faithfully reflecting the distribution of real-world images. Our super-resolution network can therefore be trained with direct pixel-wise supervision in the high resolution domain, while robustly generalizing to real input. We demonstrate the effectiveness of our approach in quantitative and qualitative experiments.