On training deep networks for satellite image super-resolution
This work addresses the problem of improving SRR for satellite image analysis, which is incremental as it focuses on optimizing data preparation rather than introducing a new model.
The paper investigates how the method of generating low-resolution training data affects super-resolution reconstruction (SRR) performance for satellite images, finding that the widely-used bicubic downsampling approach is not the most effective and that training data characteristics significantly impact accuracy.
The capabilities of super-resolution reconstruction (SRR)---techniques for enhancing image spatial resolution---have been recently improved significantly by the use of deep convolutional neural networks. Commonly, such networks are learned using huge training sets composed of original images alongside their low-resolution counterparts, obtained with bicubic downsampling. In this paper, we investigate how the SRR performance is influenced by the way such low-resolution training data are obtained, which has not been explored up to date. Our extensive experimental study indicates that the training data characteristics have a large impact on the reconstruction accuracy, and the widely-adopted approach is not the most effective for dealing with satellite images. Overall, we argue that developing better training data preparation routines may be pivotal in making SRR suitable for real-world applications.