基于跨域小样本学习的SAR图像目标识别方法

doi:10.11896/jsjkx.230800136

摘要/Abstract

摘要： 由于SAR图像获取难度大,可供研究的样本数量较少,解决有限样本条件下SAR图像目标识别问题成为业界公认的挑战。随着深度学习在计算机视觉领域的发展,衍生出了多种小样本图像分类方法,因此考虑采用跨域小样本学习范式解决小样本SAR图像目标识别问题。具体地,先在多个源域中训练得到不同域的特征提取器,而后通过知识蒸馏的方法获取一个通用的特征提取器,这里采用中心核对齐的方法,将提取的特征映射到一个更高维的空间,从而更好地区分原特征之间的非线性相似性;通过上一阶段获得的通用特征提取器提取目标域图像特征,最后采用原型网络的方法预测样本的类别。实验证明,该方法在缩减模型参数的同时,获得了88.61%的准确率,为解决小样本SAR图像目标识别问题提供了新的思路。

关键词: 深度学习, 元学习, 跨域小样本学习, SAR图像目标识别, 知识蒸馏

Abstract: Due to the difficulty in acquiring SAR images and the scarce number of samples available for research,solving the SAR image target recognition problem under few shot conditions has become a community-recognized challenge.With the development of deep learning in the field of computer vision,a variety of few-shot image classification methods have been derived,so a cross-domain few-shot learning paradigm is considered to solve the few-shot SAR image target recognition problem.Concretely,the feature extractors of different domains are first trained inmultiple source domains,while a generalized feature extractor is obtained by knowledge distillation.In this stage,the central kernet alignment method is used to map the extracted features to a higher dimensional space,so as to better distinguish the nonlinear similarity between the original features.Then the target domain image features are extracted by the generalized feature extractor obtained in the previous stage.Finally,a prototype network approach to predict the class of the sample.The experiment proves that the method obtains 88.61% accuracy while reducing the model parameters,which provides a new method for solving the target recognition problem of SAR images with scarce samples.

Key words: Deep learning, Meta learning, Cross domain few shot learning, SAR image target recognition, Knowledge distillation

中图分类号:

TP18

史松昊, 王晓丹, 杨春晓, 王艺菲. 基于跨域小样本学习的SAR图像目标识别方法[J]. 计算机科学, 2024, 51(6A): 230800136-7. https://doi.org/10.11896/jsjkx.230800136

SHI Songhao, WANG Xiaodan, YANG Chunxiao, WANG Yifei. SAR Image Target Recognition Based on Cross Domain Few Shot Learning[J]. Computer Science, 2024, 51(6A): 230800136-7. https://doi.org/10.11896/jsjkx.230800136

参考文献

[1]YING Z L,WANG W Q,XU Y,et al.Twin Self-SupervisedLearning Method for Small Sample SAR Images Automatic Target Recognition[J/OL].Signal Processing:1-13.[2023-07-13].http://kns.cnki.net/kcms/detail/11.2406.TN.20230321.1926.010.html.
[2]FENG B D,YANG H T,WANG J N,et al.SAR Image TargetRecognition Algorithm Based on Data Fusion[J].Computer System Applications,2022,31(12):342-349.
[3]WANG Y Y.SAR TargetRecognition Based on Modified Sparse Representation[J/OL].Electronics Optics & Control:1-7.[2023-08-21].http://kns.cnki.net/kcms/detail/41.1227.TN.20230727.1127.004.html.
[4]KANG Z Q,ZHANG S Q,FENG S J,et al.Sparse Prior-Guided CNN Learning for SAR Images Target Recognition[J].Journal Of Signal Processing,2023,39(4):737-750.
[5]TANG H,LI Z,PENG Z,et al.Blockmix:meta regularization and self-calibrated inference for metric-based meta-learning[C]//Proceedings of the 28th ACM International Conference on Multimedia.2020:610-618.
[6]ANTONIOU A,EDWARDS H,STORKEY A.How to trainyour MAML[J].arXiv:1810.09502,2018.
[7]SUNG F,YANG Y,ZHANG L,et al.Learning to compare:Relation network for few-shot learning[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2018:1199-1208.
[8]XIE Y,FU Y,TAI Y,et al.Learning To Memorize Feature Hallucination for One-Shot Image Generation[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.2022:9130-9139.
[9]SCHROEDER B,CUI Y.Fgvcx fungi classification challenge 2018[J/OL].http://github.com/visipedia/fgvcx_fungi_comp.
[10]FINN C,ABBEEL P,LEVINE S.Model-agnostic meta-learning for fast adaptation of deep networks[C]//International Confe-rence on Machine Learning.PMLR,2017:1126-1135.
[11]SNELL J,SWERSKY K,ZEMEL R.Prototypical networks for few-shot learning[C]//Proceedings of the 31st International Conference on Neural Information Processing Systems.2017:4080-4090.
[12]CHEN W Y,LIU Y C,KIRA Z,et al.A closer look at few-shotclassification[J].arXiv:1904.04232,2019.
[13]REQUEIMA J,GORDON J,BRONSKILL J,et al.Fast andflexible multi-task classification using conditional neural adaptive processes[C]//Proceedings of 33rd International Confe-rence onNeural Information Processing Systems.2019:7959-7970.
[14]BATENI P,GOYAL R,MASRANI V,et al.Improved few-shot visual classification[C]//Proceedings of the IEEE/CVF Confe-rence on Computer Vision and Pattern Recognition.2020:14493-14502.
[15]PEREZ E,STRUB F,DE VRIES H,et al.Film:Visual reaso-ning with a general conditioning layer[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2018.
[16]DVORNIK N,SCHMID C,MAIRAL J.Selecting relevant features from a multi-domain representation for few-shot classification[C]//Computer Vision－ECCV 2020:16th European Confe-rence,Glasgow,UK,Part X 16.Springer International Publi-shing,2020:769-786.
[17]LIU L,HAMILTON W,LONG G,et al.A universal representation transformer layer for few-shot image classification[J].ar-Xiv:2006.11702,2020.
[18]PHOO C P,HARIHARAN B.Self-training for few-shot transfer across extreme task differences[J].arXiv:2010.07734,2020.
[19]FU Y,XIE Y,FU Y,et al.Me-d2n:Multi-expert domain decompositional network for cross-domain few-shot learning[C]//Proceedings of the 30th ACM International Conference on Multimedia.2022:6609-6617.
[20]LI W H,LIU X,BILEN H.Universal representation learningfrom multiple domains for few-shot classification[C]//Procee-dings of the IEEE/CVF International Conference on Computer Vision.2021:9526-9535.
[21]GUO Y,CODELLA N C,KARLINSKY L,et al.A broaderstudy of cross-domain few-shot learning[C]//Computer Vision-ECCV 2020:16th European Conference,Glasgow,UK,Part XXVII 16.Springer International Publishing,2020:124-141.
[22]YANG Y,ZHU W G,QIU L L,et al.A Survey of Research on the Target Recognition via Limited SAR Sample Based on Transfer Learning.[J/OL].Electronics Optics & Control:1-8.[2023-07-13].http://kns.cnki.net/kcms/detail/41.1227.TN.20230327.1159.002.html.
[23]CHEN W Y,LIU Y C,KIRA Z,et al.A closer look at few-shotclassification[J].arXiv:1904.04232,2019.
[24]KENDALL A,GAL Y,CIPOLLA R.Multi-task learning using uncertainty to weigh losses for scene geometry and semantics[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2018:7482-7491.
[25]ZACARIAS A,ALEXANDRE L A.Sena-cnn:Overcoming catastrophic forgetting in convolutional neural networks by selective network augmentation[C]//IAPR Workshop on Artificial Neural Networks in Pattern Recognition.Cham:Springer International Publishing,2018:102-112.
[26]ROY D,PANDA P,ROY K.Tree-CNN:a hierarchical deep convolutional neural network for incremental learning[J].Neural Networks,2020,121:148-160.
[27]XU H,ZHI S,SUN S,et al.Deep Learning for Cross-Domain Few-Shot Visual Recognition:A Survey[J].arXiv:2303.08557,2023.
[28]HINTON G,VINYALS O,DEAN J.Distilling the knowledge in a neural network[J].arXiv:1503.02531,2015.
[29]ROMERO A,BALLAS N,KAHOU S E,et al.Fitnets:Hintsfor thin deep nets[J].arXiv:1412.6550,2014.
[30]LI W H,BILEN H.Knowledge distillation for multi-task lear-ning[C]//Computer Vision-ECCV 2020 Workshops:Glasgow,UK,Part VI 16.Springer International Publishing,2020:163-176.
[31]KORNBLITH S,NOROUZI M,LEE H,et al.Similarity of neural network representations revisited[C]//International Confe-rence on Machine Learning.PMLR,2019:3519-3529.
[32]NGUYEN T,RAGHU M,KORNBLITH S.Do wide and deep networks learn the same things? uncovering how neural network representations vary with width and depth[J].arXiv:2010.15327,2020.
[33]TRIANTAFILLOU E,ZHU T,DUMOULIN V,et al.Meta-dataset:A dataset of datasets for learning to learn from few examples[J].arXiv:1903.03096,2019.
[34]RUSSAKOVSKY O,DENG J,SU H,et al.Imagenet large scale visual recognition challenge[J].International Journal of Computer Vision,2015,115:211-252.
[35]LAKE B M,SALAKHUTDINOV R,TENENBAUM J B.Human-level concept learning through probabilistic program induction[J].Science,2015,350(6266):1332-1338.
[36]MAJI S,RAHTU E,KANNALA J,et al.Fine-grained visualclassification of aircraft[J].arXiv:1306.5151,2013.
[37]WAH C,BRANSON S,WELINDER P,et al.The caltech-ucsd birds-200-2011 dataset[J/OL].https://www.docin.com/p-1472255882.html.
[38]CIMPOI M,MAJI S,KOKKINOS I,et al.Describing textures in the wild[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2014:3606-3613.
[39]JONGEJAN J,ROWLEY H,KAWASHIMA T,et al.The quick,draw!－AI experiment[J].Mount View,2016,17(2018):4.
[40]HOUBEN S,STALLKAMP J,SALMEN J,et al.Detection of traffic signs in real-world images:The German Traffic Sign Detection Benchmark[C]//The 2013 International Joint Confe-rence on Neural Networks(IJCNN).IEEE,2013:1-8.
[41]LIN T Y,MAIRE M,BELONGIE S,et al.Microsoft coco:Common objects in context[C]//Computer Vision－ECCV 2014:13th European Conference,Zurich,Switzerland,Part V 13.Springer International Publishing,2014:740-755.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed