The fully executable procedure of the U-Net model combined with the Multi-textRG algorithm to achieve fine ice-water classification.

• Main Author: Yan, Sun
• Format: Other/Unknown Material
• Language: unknown
• Published: Zenodo 2024
• Subjects: Sea ice
• Online Access: https://doi.org/10.5281/zenodo.10973107

This code source is related to the manuscript titled "Combining the U-Net model and a Multi-textRG algorithm for fine SAR ice-water classification", which will be submitted to the journal---The Cryosphere. This code source is developed based on the U-Net CNN model (coppyed from https://platform.ai4eo.eu/auto-ice/data) and a Multi-textRG algorithm proposed by us to achieve the Arctic sea-ice and water classification in high precision and with great performance for depicting ice edge details, under various conditions covering both summer and winter seasons. 1) "codes.zip" includes five code folders. 1_AI4ArcticSeaIceChallenge-U-Net ---> need to be run in Pycharm platform. 2_SAR_denoise; 3_glcm_textures_SAR_dual_polarizations; 4_SAR_Multi-textRG_algorithm; 5_newSIC_labels ---> need to be run in MATLAB. 2) "data.zip" includes the testing files. --- > there are "data_supports" and "example_processings" folders. the processing results wil be saved in "example_processings" folder. 3) The "ready-to-train-fused_01.zip" to "ready-to-train-fused_10.zip" include 200 scenes of data-fused SIC labels. They are used for further training or model experiments. Do not need to download them at the first. Using the "codes.zip" and "data.zip" can help you understand the method usage. The "ready-to-train-fused_11.zip" to "ready-to-train-fused_21.zip" include another 332 scenes of data-fused SIC labels accessible with doi:10.5281/zenodo.10974340, https://zenodo.org/records/10974340. Please feel free to download and test the method, and to give your valuable comments.