DFC30

We selected (i) the Copernicus GLO-30 DEM (COP30) dataset as the low-resolution DSM source, (ii) the Forest And Buildings removed Copernicus DEM (FABDEM) dataset as the reference DTM source, and (iii) the GRSS Data Fusion Contest 2022 (DFC2022 or grss_dfc_2022) dataset as the high-resolution image, DTM and land use mask sources to build a dataset for real-world DEM SR problems, denoted here as DFC30 (DFC2022 + FABDEM + COP30, low-resolution at ~30 m).

我们选取了以下三类源数据以构建面向真实场景数字高程模型（Digital Elevation Model, DEM）超分辨率（Super Resolution, SR）任务的数据集：(i) 哥白尼全球30米数字高程模型（Copernicus GLO-30 DEM, COP30）数据集，作为低分辨率数字表面模型（Digital Surface Model, DSM）源；(ii) 去除森林与建筑的哥白尼数字高程模型（Forest And Buildings removed Copernicus DEM, FABDEM）数据集，作为参考数字地形模型（Digital Terrain Model, DTM）源；(iii) GRSS 2022年数据融合竞赛（GRSS Data Fusion Contest 2022, DFC2022，亦称grss_dfc_2022）数据集，作为高分辨率影像、数字地形模型（DTM）与土地利用掩膜的源数据。该数据集被命名为DFC30（由DFC2022、FABDEM与COP30组成，为分辨率约30米的低分辨率版本）。