PhySAR-Seg
收藏资源简介:
PhySAR-Seg数据集是由西北工业大学自动化学院脑与人工智能实验室创建的,旨在解决极化合成孔径雷达(PolSAR)图像分割中的挑战。该数据集通过保留PolSAR图像中的物理散射机制,确保了数据的物理可解释性和信息完整性,同时提供了用户友好且存储高效的数据格式。数据集的内容包括丰富的散射特征和语义信息,适用于地形分割等应用。创建过程中,采用了微波视觉数据(MVD)产品,通过无监督的GD-Wishart分类和散射机制的重新聚类生成。该数据集的应用领域主要集中在PolSAR图像的分割任务,旨在提高分割的准确性和效率。
The PhySAR-Seg dataset was developed by the Brain and Artificial Intelligence Laboratory, School of Automation, Northwestern Polytechnical University, to address the challenges in polarimetric synthetic aperture radar (PolSAR) image segmentation. By preserving the physical scattering mechanisms inherent in PolSAR images, this dataset guarantees the physical interpretability and information integrity of the data, while providing a user-friendly and storage-efficient data format. The dataset contains abundant scattering features and semantic information, making it applicable to tasks such as terrain segmentation. During its construction, the dataset was generated using microwave vision data (MVD) products via unsupervised GD-Wishart classification and re-clustering of scattering mechanisms. Its application scenarios mainly center on PolSAR image segmentation tasks, with the goal of enhancing the accuracy and efficiency of segmentation outcomes.

- 1PolSAM: Polarimetric Scattering Mechanism Informed Segment Anything Model西北工业大学自动化学院脑与人工智能实验室 · 2024年



