卷积神经网络(Convolutional Neural Network, CNN)当前作为光学图像目标识别的主流算法，但是，由于合成孔径雷达(Synthetic Aperture Radar, SAR) 成像的复杂性，导致标注数据稀缺、标注困难，训练样本量难以达到CNN的训练样本需求。利用仿真/光学图像训练模型，再通过无监督域自适应(Unsupervised Domain Adaptation, UDA)技术弥合仿真/光学图像与真实SAR图像之间的域差异，成为一种可行路径。然而，现有UDA方法大多假设源域特征全部可迁移，忽略了部分领域特定特征可能引发负迁移，同时伪标签噪声也会影响域对齐效果。针对上述问题，本文提出一种结合特征解耦与标签噪声抑制的域自适应框架。该方法通过特征解耦模块将样本表示分离为通用特征和领域特定特征，仅对齐通用特征以降低负迁移风险；同时，提出加权广义交叉熵损失函数(Weighted Generalized Cross Entropy, WGCE)，缓解伪标签迭代过程中产生的标签噪声干扰。实验结果表明，所提方法在跨域SAR目标识别任务中有效提升了识别精度与域适应鲁棒性。
The Convolutional Neural Network (CNN) is currently the mainstream algorithm for optical image target recognition. However, due to the complexity of Synthetic Aperture Radar (SAR) imaging, annotated data are scarce and difficult to obtain, making it challenging to meet the training sample requirements of CNNs. Training models with simulated/optical images and then bridging the domain gap between simulated/optical images and real SAR images through Unsupervised Domain Adaptation (UDA) technology has emerged as a feasible solution. Nevertheless, most existing UDA methods assume that all source domain features are transferable, overlooking the fact that certain domain-specific features may cause negative transfer, while pseudo-label noise can also adversely affect domain alignment. To address the above issues, this paper proposes a domain adaptation framework that integrates feature decoupling and label noise suppression. The method employs a feature decoupling module to separate sample representations into domain-shared features and domain-specific features, aligning only the shared features to reduce the risk of negative transfer. Additionally, a weighted generalized cross-entropy loss function is introduced to mitigate the interference of label noise generated during pseudo-label iteration. Experimental results demonstrate that the proposed method effectively improves recognition accuracy and domain adaptation robustness in cross-domain SAR target recognition tasks.
[1]Zhang J, Liu J, Pan B, et al.Domain adaptation based on correlation subspace dynamic distribution alignment for remote sensing image scene classification[J].IEEE Transactions on Geoscience and Remote Sensing, 2020, 58(11):7920-7930
[2]吕小玲, 仇晓兰, 俞文明, 等.基于无监督域适应的仿真辅助目标分类方法及模型可解释性分析[J].雷达学报, 2022, 11(1):16-182
[3]LYU Xiaoling, QIU Xiaolan, YU Wenming, et al.Simulation-assisted SAR target classification based on unsupervised domain adaptation and model interpretability analysis[J].Journal of Radars, 2022, 11(1):168-182
[4]TIRANDAZ Z, AKBARIZADEH G.A two-phase algorithm based on kurtosis curvelet energy and unsupervised spectral regression for segmentation of SAR images[J].IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2016, 9(3):1244-1264
[5]LIN H P, SONG S L, YANG J A.Ship classification based on MSHOG feature and task-driven dictionary learning with structured incoherent constraints in SAR images[J].Remote Sensing, 2018, 10(2):190-
[6]胡思茹, 马福民, 秦天奇, 等.基于多特征组合的红外舰船目标识别技术[J].舰船电子工程, 2022, 42(2):185-189
[7]HU Siru, MA Fumin, QIN Tianqi, et al.Infrared ship target recognition technology based on multi feature combination[J].Ship Electronic Engineering, 2022, 42(2):185-189
[8]OUCHI K.Current status on vessel detection and classification by synthetic aperture radar for maritime security and safety[C]. 38th Symposium on Remote Sensing for Environmental Sciences, Gamagori, Japan, 2016: 3–5.
[9]ZHANG Xin, HUO Chunlei, XU Nuo, et al.Multitask learning for ship detection from synthetic aperture radar images[J].IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2021, 14:8048-8062
[10]张翼鹏, 卢东东, 仇晓兰, 等.基于散射点拓扑和双分支卷积神经网络的图像小样本舰船分类[J].雷达学报中英文, 2024, 13(2):411-427
[11]HANG Yipeng, LU Dongdong, QIU Xiaolan, et al.Few-shot ship classification of SAR images via scattering point topology and dual-branch convolutional neural network[J].Journal of Radars, 2024, 13(2):411-427
[12]Wang H, Chen S, Xu F, et al.Application of deep-learning algorithms to MSTAR data[C]//2015 IEEE International Geoscience and Remote Sensing Symposium (IGARSS). IEEE, 2015: 3743-3745.
[13]Huang Z, Datcu M, Pan Z, et al.Deep SAR-Net: Learning objects from signals[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2020, 161: 179-193.
[14]Chen J, Qiu X, Ding C, et al.SAR image classification based on spiking neural network through spike-time dependent plasticity and gradient descent[J]. ISPRS journal of photogrammetry and remote sensing, 2022, 188: 109-124.
[15]Li Y, Du L, Wei D.Multiscale CNN based on component analysis for SAR ATR[J]. IEEE Transactions on Geoscience and Remote Sensing, 2021, 60: 1-12.
[16]Shi Y, Du L, Li C, et al.Unsupervised domain adaptation for SAR target classification based on domain-and class-level alignment: From simulated to real data[J]. ISPRS journal of photogrammetry and remote sensing, 2024, 207: 1-13.
[17]Long M, Zhu H, Wang J, et al.Deep transfer learning with joint adaptation networks[C]//International conference on machine learning. PMLR, 2017: 2208-2217.
[18]Zhu Y, Zhuang F, Wang J, et al.Deep subdomain adaptation network for image classification[J].IEEE transactions on neural networks and learning systems, 2020, 32(4):1713-1722
[19]Long M, Cao Z, Wang J, et al.Conditional adversarial domain adaptation[J]. Advances in neural information processing systems, 2018, 31.
[20]Du Z, Li J, Su H, et al.Cross-domain gradient discrepancy minimization for unsupervised domain adaptation [C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2021: 3937-3946.
[21]Ghifary M, Kleijn W B, Zhang M, et al.Deep reconstruction-classification networks for unsupervised domain adaptation[C]//European conference on computer vision. Cham: Springer International Publishing, 2016: 597-613.
[22]Bousmalis K, Trigeorgis G, Silberman N, et al.Domain separation networks[J]. Advances in neural information processing systems, 2016, 29.
[23]Sankaranarayanan S, Balaji Y, Castillo C D, et al.Generate to adapt: Aligning domains using generative adversarial networks[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2018: 8503-8512.
[24]Lewis B, Scarnati T, Sudkamp E, et al.A SAR dataset for ATR development: the Synthetic and Measured Paired Labeled Experiment (SAMPLE)[C]//Algorithms for Synthetic Aperture Radar Imagery XXVI. SPIE, 2019, 10987: 39-54.
[25]Shi Y, Du L, Guo Y, et al.Unsupervised domain adaptation for ship classification via progressive feature alignment: From optical to SAR images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2024.
[26]Hou X, Ao W, Song Q, et al.FUSAR-Ship: Building a high-resolution SAR-AIS matchup dataset of Gaofen-3 for ship detection and recognition[J].Science China Information Sciences, 2020, 63(4):140303-
[27]Ganin Y, Lempitsky V.Unsupervised domain adaptation by backpropagation[C]//International conference on machine learning. PMLR, 2015: 1180-1189.
[28]Sharma A, Kalluri T, Chandraker M.Instance level affinity-based transfer for unsupervised domain adaptation[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2021: 5361-5371.
[29]Sun B, Saenko K.Deep coral: Correlation alignment for deep domain adaptation[C]//European conference on computer vision. Cham: Springer International Publishing, 2016: 443-450.
[30]Zhu Y, Zhuang F, Wang J, et al.Multi-representation adaptation network for cross-domain image classification[J]. Neural Networks, 2019, 119: 214-221.
[31]Dai S, Cheng Y, Zhang Y, et al.Contrastively smoothed class alignment for unsupervised domain adaptation [C]//Proceedings of the Asian conference on computer vision. 2020.
CJA