当前位置: X-MOL 学术IEEE Geosci. Remote Sens. Lett. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Exploring PolSAR Images Representation via Self-Supervised Learning and Its Application on Few-Shot Classification
IEEE Geoscience and Remote Sensing Letters ( IF 4.8 ) Pub Date : 2022-08-11 , DOI: 10.1109/lgrs.2022.3198135
Wu Zhang 1 , Zongxu Pan 1 , Yuxin Hu 1
Affiliation  

Deep learning methods have attracted much attention in the field of polarimetric synthetic aperture radar (PolSAR) image classification over the past few years. However, for supervised learning (SL)-based methods, it is quite difficult to get large amounts of high-quality and labeled PolSAR data in real applications. In addition, there is a problem of poor generalization for the method of specific supervision labels. To solve the abovementioned issue, we explore how to learn representations from unlabeled data from a new perspective. In this letter, a self-supervised PolSAR representation learning (SSPRL) method is proposed. Different from the SL-based methods, SSPRL aims to learn PolSAR image representations via an unsupervised learning approach. Specifically, a self-SL (SSL) method without negative samples is explored, and a positive sample generation (PSG) approach and a novel encoder architecture designed for PolSAR images are proposed. Moreover, mix-up is implemented as a regularization strategy. Furthermore, the convolutional encoder is utilized to transfer the feature representation from the unlabeled PolSAR data to the downstream task, that is, to achieve the few-shot PolSAR classification. Comparative experimental results on two widely used PolSAR benchmark datasets verify the effectiveness of the proposed method and demonstrate that the SSPRL produces impressive performance on few-shot classification task compared with state-of-the-art algorithms.

中文翻译:

通过自监督学习探索 PolSAR 图像表示及其在 Few-Shot 分类中的应用

近年来,深度学习方法在极化合成孔径雷达(PolSAR)图像分类领域备受关注。然而,对于基于监​​督学习(SL)的方法,在实际应用中很难获得大量高质量和标记的 PolSAR 数据。此外,具体监督标签的方法存在泛化性差的问题。为了解决上述问题,我们从一个新的角度探索如何从未标记的数据中学习表示。在这封信中,提出了一种自我监督的 PolSAR 表示学习 (SSPRL) 方法。与基于 SL 的方法不同,SSPRL 旨在通过无监督学习方法学习 PolSAR 图像表示。具体来说,探索了一种没有负样本的self-SL(SSL)方法,并提出了一种为 PolSAR 图像设计的正样本生成 (PSG) 方法和一种新颖的编码器架构。此外,混合是作为正则化策略实施的。此外,卷积编码器用于将特征表示从未标记的 PolSAR 数据传输到下游任务,即实现少样本 PolSAR 分类。两个广泛使用的 PolSAR 基准数据集的比较实验结果验证了所提出方法的有效性,并证明与最先进的算法相比,SSPRL 在少样本分类任务上产生了令人印象深刻的性能。利用卷积编码器将特征表示从未标记的 PolSAR 数据传递到下游任务,即实现少样本 PolSAR 分类。两个广泛使用的 PolSAR 基准数据集的比较实验结果验证了所提出方法的有效性,并证明与最先进的算法相比,SSPRL 在少样本分类任务上产生了令人印象深刻的性能。利用卷积编码器将特征表示从未标记的 PolSAR 数据传递到下游任务,即实现少样本 PolSAR 分类。两个广泛使用的 PolSAR 基准数据集的比较实验结果验证了所提出方法的有效性,并证明与最先进的算法相比,SSPRL 在少样本分类任务上产生了令人印象深刻的性能。
更新日期:2022-08-11
down
wechat
bug