基于深度生成网络的雷达HRRP生成技术
|
摘要
雷达数据生成在目标识别等任务中发挥重要的作用。现有雷达数据生成方法包括电磁仿真、视线追踪等,存在对模型误差敏感、计算量大等问题。本文面向雷达HRRP(high resolution range profile)数据提出一种基于深度生成网络的雷达数据生成方法,在模型先验信息未知的情况下,由雷达HRRP数据集训练得到深度生成网络,从而实现雷达HRRP数据的快速生成。实测数据处理结果表明该方法生成HRRP与数据集中HRRP极为相似,生成HRRP可以应用于增强雷达HRRP数据集、改善数据不平衡问题等。
Radar data generation plays an important role in radar applications, e.g. radar target recognition. Radar data generation method contains simulation based on statistic model and electromagnetic simulation. These methods are sensitivity to model error, and the electromagnetic simulation always faces heavy calculation. In this paper, a method based on deep generative model is proposed in which a generative model can be trained with only few data samples, and radar data can be generated rapidly without heavy calculation. This method was applied to generate radar HRRP, and the result shows that target HRRP can be generated, and the generated HRRPs are similar to real radar data in visual and in statistic domain, and the generated HRRP can be used to eliminate the effect of imbalance problems.
Radar data generation plays an important role in radar applications, e.g. radar target recognition. Radar data generation method contains simulation based on statistic model and electromagnetic simulation. These methods are sensitivity to model error, and the electromagnetic simulation always faces heavy calculation. In this paper, a method based on deep generative model is proposed in which a generative model can be trained with only few data samples, and radar data can be generated rapidly without heavy calculation. This method was applied to generate radar HRRP, and the result shows that target HRRP can be generated, and the generated HRRPs are similar to real radar data in visual and in statistic domain, and the generated HRRP can be used to eliminate the effect of imbalance problems.
引文
[1] Zhan Ying,Hu Du,Wang Yuntao,et al.Semisupervised hyperspectral image classification based on generative adversarial networks[J].IEEE Geoscience Remote Sensing Letters,2018,15(2):212-216.
[2] Skolnik M I.Introduction to radar systems[J].McGraw-Hill Electrical Engineering Series,2001(2):1-7.
[3] Franceschetti G,Migliaccio M,Riccio D.The SAR simulation:An overview[J].Geoscience Remote Sensing,1995,3(3):2283-2285.
[4] Balz T,Hammer H,Auer S.Potentials and limitations of SAR image simulators-A comparative study of three simulation approaches[J].Remote Sensing,2015,101(101):102-109.
[5] Ulaby F T,Moore R K,Fung A K.Microwave remote sensing[J].Theory to Applications,1986.vol.III.
[6] Xu Feng,Jin Yaqiu.Imaging simulation of polarimetric SAR for a comprehensive terrain scene using the mapping and projection algorithm[J].IEEE Transaction on Geoscience Remote Sensing,2006,44(11):3219-3234.
[7] Brunner D,Lemoine G,Greidanus H,et al.Radar imaging simulation for urban structures[J].IEEE Geoscience Remote Sensing Letters,2010,8(1):68-72.
[8] Hammer H,Schulz K.Dedicated SAR simulation tools for ATR and scene analysis[J].Proceedings of SPIE,2011,8179:81790N.
[9] Balz T,Stilla U.Hybrid GPU-based single- and double-bounceSAR simulation[J].IEEE Transaction Geoscience Remote Sensing,2009,47(10):3519-3529.
[10] Hinton G E,Salakhutdinov R R.Reducing the dimensionality of data with neural networks[J].Science,2006,313(5786):504-507.
[11] Kingma D P.Fast gradient-based inference with continuous latent variable models in auxiliary form[J].Computer Science,2013.
[12] Goodfellow I,et al.Generative adversarial nets[J].Neural Information Processing Systems,2014:2672-2680.
[13] Radford,Metz L,Chintala S.Unsupervised representation learning with deep convolutional generative adversarial networks[J].Computer Science,2015.
[2] Skolnik M I.Introduction to radar systems[J].McGraw-Hill Electrical Engineering Series,2001(2):1-7.
[3] Franceschetti G,Migliaccio M,Riccio D.The SAR simulation:An overview[J].Geoscience Remote Sensing,1995,3(3):2283-2285.
[4] Balz T,Hammer H,Auer S.Potentials and limitations of SAR image simulators-A comparative study of three simulation approaches[J].Remote Sensing,2015,101(101):102-109.
[5] Ulaby F T,Moore R K,Fung A K.Microwave remote sensing[J].Theory to Applications,1986.vol.III.
[6] Xu Feng,Jin Yaqiu.Imaging simulation of polarimetric SAR for a comprehensive terrain scene using the mapping and projection algorithm[J].IEEE Transaction on Geoscience Remote Sensing,2006,44(11):3219-3234.
[7] Brunner D,Lemoine G,Greidanus H,et al.Radar imaging simulation for urban structures[J].IEEE Geoscience Remote Sensing Letters,2010,8(1):68-72.
[8] Hammer H,Schulz K.Dedicated SAR simulation tools for ATR and scene analysis[J].Proceedings of SPIE,2011,8179:81790N.
[9] Balz T,Stilla U.Hybrid GPU-based single- and double-bounceSAR simulation[J].IEEE Transaction Geoscience Remote Sensing,2009,47(10):3519-3529.
[10] Hinton G E,Salakhutdinov R R.Reducing the dimensionality of data with neural networks[J].Science,2006,313(5786):504-507.
[11] Kingma D P.Fast gradient-based inference with continuous latent variable models in auxiliary form[J].Computer Science,2013.
[12] Goodfellow I,et al.Generative adversarial nets[J].Neural Information Processing Systems,2014:2672-2680.
[13] Radford,Metz L,Chintala S.Unsupervised representation learning with deep convolutional generative adversarial networks[J].Computer Science,2015.