Feature enhanced deep learning network for digital elevation model super-resolution

High-resolution digital elevation model (HR DEM) plays an important role in hydrological analysis, cartographic generalization, and national security. As the main high-precision DEM data supplementary method, super-resolution (DEM SR) based on deep learning has been widely studied. However, its accuracy fallen into a bottleneck at present, which is more prominent complex regions. The reason for this issue that existing methods are difficult to capture enough local features from low-resolution (LR) input data, part of global information (Contour long-distance such as rivers ridges) will also be lost network transmission process. To resolve issue, novel feature-enhanced (FEN) designed paper. proposed FEN includes feature SR (GFSR) module (LFSR) module. former provides by using interpolation method (Kriging) including geographical laws (spatial autocorrelation); latter fully captures integrating powerful extraction modules then sufficient tasks. Thus, tasks regions can realized results GFSR LFSR modules. Extensive experiments show achieves state-of-the-art performance facing Specifically, compared with (TfaSR, SRResNet, Bicubic, SRCNN, Kriging), result closer HR retain features. Meanwhile, than 20% ahead other accuracy.