Data and Code for “A Hausdorff-Guided Deep Learning Approach for Monitoring the Motion of Rotating Arctic Ice Floes”
Description
Data for the manuscript entitled “A Hausdorff-Guided Deep Learning Approach for Monitoring the Motion of Rotating Arctic Ice Floes.” This dataset is specifically curated to support the submission and reproducibility of the above-mentioned manuscript. It provides a comprehensive collection of resources covering the full experimental pipeline, including source data, intermediate processing results, comparative experiments, final outputs, and implementation code. The repository contains both data and code: Dataset for HDL-IFM This archive contains all data used in the study and is further organized into five subfolders, which systematically document the workflow from raw data to experimental results: Source Data This folder contains 226 multi-temporal Arctic ice floe images used as training data for the deep learning framework proposed in the manuscript. Ice Floe Data for the Experiment This folder includes both the original ice floe imagery and the preprocessed data used in the experiments, reflecting the data preparation procedures applied in the study. Extraction and Matching Results of Traditional Methods for Ice Floe Monitoring This folder presents the feature extraction and matching results obtained using traditional methods, including SIFT and A-KAZE, which serve as baseline comparisons. Experimental Results This folder contains the motion monitoring results of ice floes between consecutive days, as well as vector maps illustrating short-term continuous motion trajectories. Influence of Ice Floe Rotation on Deep-Model Matching This folder includes spatial visualization results under controlled rotations (72 rotations with a 5° interval), along with quantitative analyses demonstrating the impact of rotation on matching accuracy. Code for HDL-IFM This folder contains the implementation of the proposed Hausdorff-guided deep learning framework (HDL-IFM), including scripts for data preprocessing, model training, feature extraction, matching, and evaluation. The code is organized to facilitate reproducibility of all experiments reported in the manuscript. This dataset and code repository jointly support the development, validation, and comparative analysis of the proposed framework for detecting feature points and estimating the motion of rotating ice floes under complex Arctic marginal ice zone conditions.
Files
Steps to reproduce
The Sentinel-2 Level-2 surface reflectance products provided by the European Space Agency (ESA) serve as the primary data source for extracting ice floe distribution and motion information. Only clear-sky imagery with minimal cloud cover is selected to ensure data quality. True-colour composites are generated using Band 2 (blue, 458–523 nm), Band 3 (green, 543–578 nm), and Band 4 (red, 650–680 nm), each with a spatial resolution of 10 m. To enable accurate area calculations, the projected coordinate system is converted from WGS 1984 UTM Zone 47N to the North Pole Lambert azimuthal equal-area projection. Ice–water classification is performed using a dynamic threshold determined by the local minimum between the two peaks in the grayscale histogram of each image, resulting in a binary mask where white pixels represent ice and black pixels represent water. Small or spurious ice floes with blurred boundaries are removed through a morphological “erosion–marking–dilation” procedure, yielding independent ice floes for further analysis. For deep learning model training, 226 representative ice floe images acquired between 2017 and 2023 are selected. Paired images are generated via affine and perspective transformations, and pseudo-labels are created using homographic adaptation, following a self-supervised strategy. The SuperPoint model is trained with the Adam optimizer (initial learning rate = 1×10⁻⁴, batch size = 8) for 20,000 iterations, with checkpoints saved every 2,000 iterations. The SuperGlue model is subsequently trained end‑to‑end using the transferred SuperPoint weights, also with Adam (initial learning rate = 1×10⁻⁴, decay factor = 0.1 every 10 epochs) for up to 500 epochs (batch size = 4), employing early stopping if the validation loss does not improve for 50 consecutive epochs. All experiments are conducted on a Windows 10 machine equipped with an NVIDIA RTX 4060 GPU (8 GB memory) using CUDA 11.8 and Python 3.11.5. Keypoint extraction and feature matching are performed with a keypoint threshold of 0.001 and a match threshold of 0.1. To facilitate reproducibility, the complete code and trained models will be made publicly available at the URL provided in the paper.
Institutions
- Northwest Institute of Eco-Environment and ResourcesGansu, Lanzhou