Sentinel-1 GRD dataset

该数据集由长时间序列的哨兵-1 GRD图像创建而成，旨在用于训练深度学习模型，以过滤合成孔径雷达图像中的斑点噪声。该数据集包含经过多视处理并投射到地面范围的聚焦合成孔径雷达数据，同时采用原始功率值以保留图像中的清晰特征。数据集以多次获取的强度数据的时空平均为尺度，其任务是针对合成孔径雷达图像进行斑点噪声过滤。This dataset is constructed from long-time-series Sentinel-1 GRD images, and is specifically designed for training deep learning models to perform speckle noise filtering on synthetic aperture radar (SAR) imagery. It includes focused SAR data that has undergone multi-looking processing and been projected to ground range, while retaining raw power values to preserve distinct features in the imagery. The dataset leverages the spatio-temporal average of multiple acquired intensity measurements as the reference ground truth, with its core task being speckle noise filtering for SAR images.

搜集汇总

背景与挑战

背景概述

Sentinel-1 GRD数据集是来自欧洲Copernicus计划的C波段合成孔径雷达地面距离检测产品，覆盖2014年至2026年，重访周期6天。该数据集提供校准和地形校正后的影像，包含三种空间分辨率（10米、25米、40米）和多种极化组合（如VV、HH等），并经过热噪声去除、辐射定标等预处理，以分贝为单位表示，适用于雷达遥感分析和应用。