Super-resolution results are usually measured by full-reference image quality metrics or human rating scores. However, these evaluation methods are general image quality measurement, and do not account for the nature of the super-resolution problem. In this work, we analyze the evaluation problem based on the one-to-many mapping nature of super-resolution, and propose a novel distribution-based metric for super-resolution. Starting from the distribution distance, we derive the proposed metric to make it accessible and easy to compute. Through a human subject study on super-resolution, we show that the proposed metric is highly correlated with the human perceptual quality, and better than most existing metrics. Moreover, the proposed metric has a higher correlation with the fidelity measure compared to the perception-based metrics. To understand the properties of the proposed metric, we conduct extensive evaluation in terms of its design choices, and show that the metric is robust to its design choices. Finally, we show that the metric can be used to train super-resolution networks for better perceptual quality.
View on arXiv