基于改进ResNet网络的复数SAR图像舰船目标识别方法

doi:10.12305/j.issn.1001-506X.2022.12.08

摘要/Abstract

摘要：

合成孔径雷达(synthetic aperture radar, SAR)采用微波相干成像, 因此SAR图像本质上是复数的。传统基于神经网络的SAR图像目标识别方法, 通常只处理SAR图像的幅度信息, 无法有效利用SAR图像特有的复数信息。本文面向SAR图像中的舰船目标识别应用, 从SAR图像的本质出发, 首先通过组合SAR图像的实部、虚部和幅度三通道信息, 隐式地提供了输入数据的复数信息表示; 然后在ResNet18网络及其结构基础上引入通道注意力机制, 使网络能自适应学习实部、虚部和幅度三通道之间包含的复数信息; 最后引入标签平滑正则化, 解决因复数数据集样本较少出现的过拟合现象。基于OpenSARShip数据集的实验结果表明, 所提方法可以较好利用SAR图像本身的复数信息, 在一定程度上提升了基于深度神经网络的舰船目标识别效果。

关键词: 合成孔径雷达, 神经网络, 复数信息, 舰船目标识别

Abstract:

Synthetic aperture radar (SAR) uses microwave coherent imaging, so SAR images are complex in essence. Traditional SAR image target recognition methods based on neural networks usually only process the amplitude information of SAR images, but cannot effectively use the unique complex information of SAR images. This paper aims at the application of ship target recognition in SAR image. Starting from the essence of SAR image, the complex information representation of input data is implicitly provided by combining the real part, imaginary part and amplitude information of SAR image. Then the channel attention mechanism is introduced on the basis of the ResNet18 network and its structure, so that the network can adaptively learn the complex information contained in the three channels of real part, imaginary part and amplitude. Finally, label smoothing regularization is introduced to solve the over-fitting phenomenon due to the lack of samples in complex data sets. The experimental results based on the OpenSARShip data set show that the method proposed in this paper can make better use of the complex information of the SAR image itself, and to a certain extent improve the effect of ship target recognition based on neural network.

Key words: synthetic aperture radar (SAR), neural networks, complex information, ship target recognition

中图分类号:

TN957.51

雷禹, 冷祥光, 周晓艳, 孙忠镇, 计科峰. 基于改进ResNet网络的复数SAR图像舰船目标识别方法[J]. 系统工程与电子技术, 2022, 44(12): 3652-3660.

Yu LEI, Xiangguang LENG, Xiaoyan ZHOU, Zhongzhen SUN, Kefeng JI. Recognition method of ship target in complex SAR image based on improved ResNet network[J]. Systems Engineering and Electronics, 2022, 44(12): 3652-3660.

图/表 11

图1

图2

图3

图4

图5

图6

图7

表1

表2

表3

图8

参考文献 30

1	冷祥光, 计科峰, 熊博莅, 等. 面向舰船目标检测的单通道复值SAR图像统计建模方法研究[J]. 雷达学报, 2020, 9 (3): 477- 496.
	LENG X G , JI K F , XIONG B L , et al. Statistical modeling methods of single channel complex-valued SAR images for ship detection[J]. Journal of Radars, 2020, 9 (3): 477- 496.
2	杜兰, 王兆成, 王燕, 等. 复杂场景下单通道SAR目标检测及鉴别研究进展综述[J]. 雷达学报, 2020, 9 (1): 34- 54.
	DU L , WANG Z C , WANG Y , et al. Survey of research progress on target detection and discrimination of single-channel SAR images for complex scenes[J]. Journal of Radars, 2020, 9 (1): 34- 54.
3	雷禹, 冷祥光, 计科峰. 基于Google Earth Engine的海量舰船目标SAR图像处理应用研究[J]. 信号处理, 2021, 37 (6): 1075- 1085.
	LEI Y , LENG X G , JI K F . Research on the application of SAR image processing of massive ship targets based on Google Earth Engine[J]. Journal of Signal Processing, 2021, 37 (6): 1075- 1085.
4	张荻. 基于深度学习的SAR图像舰船目标识别方法研究[D]. 长沙: 国防科技大学, 2017.
	ZHANG D. Study on recognition of ships in SAR imagery based on deep learning methods[D]. Changsha: National University of Defense Technology, 2017.
5	康妙. 基于深度学习的SAR图像舰船目标检测与识别技术研究[D]. 长沙: 国防科技大学, 2017.
	KANG M. Study on the methods of SAR ship detection and recognition based on deep learning[D]. Changsha: National University of Defense Technology, 2017.
6	李家起, 江政杰, 姚力波, 等. 一种基于深度学习的舰船目标融合识别算法[J]. 舰船电子工程, 2020, 40 (9): 31- 35. 31-35, 171 doi: 10.3969/j.issn.1672-9730.2020.09.008
	LI J Q , JIANG Z J , YAO L B , et al. A ship target fusion recognition algorithm based on deep learning[J]. Ship Electronic Engineering, 2020, 40 (9): 31- 35. 31-35, 171 doi: 10.3969/j.issn.1672-9730.2020.09.008
7	MARGARIT G , MALLORQUI J J , FABREGAS X . Single-pass polarimetric SAR interferometry for vessel classification[J]. IEEE Trans.on Geoscience and Remote Sensing, 2007, 45 (11): 3494- 3502. doi: 10.1109/TGRS.2007.897437
8	PALADIN R , MARTORELL M , BERIZZI F . Classification of man-made targets via invariant coherency-matrix eigenvector decomposition of polarimetric SAR/ISAR images[J]. IEEE Trans.on Geoscience and Remote Sensing, 2011, 49 (8): 3022- 3034. doi: 10.1109/TGRS.2011.2116121
9	HINTON G E , SALAKHUTDINOV R R . Reducing the dimensionality of data with neural networks[J]. Science, 2006, 313 (5786): 504- 507. doi: 10.1126/science.1127647
10	邵嘉琦, 曲长文, 李健伟, 等. 基于CNN的不平衡SAR图像舰船目标识别[J]. 电光与控制, 2019, 26 (9): 90- 97. doi: 10.3969/j.issn.1671-637X.2019.09.019
	SHAO J Q , QU C W , LI J W , et al. CNN based ship target recognition of imbalanced SAR image[J]. Electronics Optics & Control, 2019, 26 (9): 90- 97. doi: 10.3969/j.issn.1671-637X.2019.09.019
11	LI L L , MA L Y , JIAO L C , et al. Complex contourlet-CNN for polarimetric SAR image classification[J]. Pattern Recognition, 2020, 100, 107110. doi: 10.1016/j.patcog.2019.107110
12	李玲玲. 量子进化优化与深度复神经网络学习算法及其应用[D]. 西安: 西安电子科技大学, 2017.
	LI L L. Quantum evolutional optimization and deep complex neural networks learning[D]. Xi'an: Xidian University, 2017.
13	LECUN Y , BENGIO Y , HINTON G . Deep learning[J]. Nature, 2015, 521 (7553): 436- 444. doi: 10.1038/nature14539
14	DONAHUE J, HENDRICKS L A, GUADARRAMD S, et al. Long-term recurrent convolutional networks for visual recognition and description[C]//Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, 2015: 2625-2634.
15	HINTON G , DENG L , YU D , et al. Deep neural networks for acoustic modeling in speech recognition: the shared views of four research groups[J]. IEEE Signal Processing Magazine, 2012, 29 (6): 82- 97. doi: 10.1109/MSP.2012.2205597
16	EL-DARYMLI K , MCGUIRE P , GILL E W , et al. Characterization and statistical modeling of phase in single-channel synthetic aperture radar imagery[J]. IEEE Trans.on Aerospace and Electronic Systems, 2015, 51 (3): 2071- 2092. doi: 10.1109/TAES.2015.140711
17	EL-DARYMLI K, MOLONEY C, GILL E, et al. Nonlinearity and the effect of detection on single-channel synthetic aperture radar imagery[C]//Proc. of the IEEE OCEANS, 2014.
18	刘国祥. SAR成像原理与图像特征[J]. 四川测绘, 2004, 27 (3): 141- 143.
	LIU G X . Principles of imaging SAR and characteristics of image[J]. Surveying and Mapping of Sichuan, 2004, 27 (3): 141- 143.
19	HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]//Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, 2016: 770-778.
20	刘家麒, 陈渤, 介茜. 基于注意力机制和双向GRU模型的雷达HRRP目标识别[J]. 雷达学报, 2019, 8 (5): 589- 597.
	LIU J Q , CHEN B , JIE X . Radar high-resolution range profile target recognition based on attention mechanism and bidirectional gated recurrent[J]. Journal of Radars, 2019, 8 (5): 589- 597.
21	赵文清, 程幸福, 赵振兵, 等. 注意力机制和Faster RCNN相结合的绝缘子识别[J]. 智能系统学报, 2020, 15 (1): 92- 98.
	ZHAO W Q , CHENG X F , ZHAO Z B , et al. Insulator recognition based on attention mechanism and Faster RCNN[J]. CAAI Transactions on Intelligent Systems, 2020, 15 (1): 92- 98.
22	应自炉, 宣晨, 翟懿奎, 等. 面向小样本SAR图像识别的自注意力多尺度特征融合网络[J]. 信号处理, 2020, 36 (11): 1846- 1858.
	YING Z L , XUAN C , ZHAI Y K , et al. Self-attention multiscale feature fusion network for small sample SAR image recognition[J]. Journal of Signal Processing, 2020, 36 (11): 1846- 1858.
23	HU J, SHEN L, SUN G. Squeeze-and-excitation networks[C]//Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, 2018: 7132-7141.
24	SZEGEDY C, VANHOUCKE V, IOFFE S, et al. Rethinking the inception architecture for computer vision[C]//Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, 2016: 2818-2826.
25	HE T, ZHANG Z, ZHANG H, et al. Bag of tricks for image classification with convolutional neural networks[C]//Proc. of the IEEE Conference on Computer Vision and Pattern Recognition, 2019: 558-567.
26	MULLER R, KORNBLITH S, HINTON G E. When does label smoothing help?[C]//Proc. of the 33rd International Conference on Neural Information Processing Systems, 2019, 422: 4694-4703.
27	HUANG L Q , LIU B , LI B Y , et al. OpenSARShip: a dataset dedicated to Sentinel-1 ship interpretation[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2017, 11 (1): 195- 208.
28	LI B Y, LIU B, HUANG L Q, et al. OpenSARShip 2.0: a large-volume dataset for deeper interpretation of ship targets in Sentinel-1 imagery[C]//Proc. of the IEEE SAR in Big Data Era: Models, Methods and Applications, 2017: 1-5.
29	赵楠, 张小芳, 张利军. 不平衡数据分类研究综述[J]. 计算机科学, 2018, 45 (B6): 22- 27. 22-27, 57
	ZHAO N , ZHANG X F , ZHANG L J . Overview of imbalanced data classification[J]. Computer Science, 2018, 45 (B6): 22- 27. 22-27, 57
30	李海涛. 基于深度学习的图像识别鲁棒性研究[D]. 南京: 南京邮电大学, 2018.
	LI H T. Research on image recognition robustness based on depth learning[D]. Nanjing: Nanjing University of Posts and Telecommunications, 2018.

超参数	参数值
初始学习率	0.000 1
动量系数	0.1
迭代次数	100
批处理数	16
线程数	8
SE-r	16
LSR-ε	0.1

网络模型	数据类型	精确率		准确率
网络模型	数据类型	货船	油船	准确率
VGG16Net	Z₁	0.727	0.730	0.727
VGG16Net	Z₂	0.722	0.736	0.730
VGG16Net	Z₃	0.732	0.768	0.739
VGG16Net	Z₄	0.736	0.709	0.722
GoogLeNet	Z₁	0.707	0.697	0.697
GoogLeNet	Z₂	0.709	0.722	0.713
GoogLeNet	Z₃	0.722	0.736	0.727
GoogLeNet	Z₄	0.729	0.691	0.713
ResNet18	Z₁	0.703	0.80	0.739
ResNet18	Z₂	0.744	0.753	0.750
ResNet18	Z₃	0.753	0.786	0.772
ResNet18	Z₄	0.760	0.750	0.759

网络模型	数据类型	精确率		准确率
网络模型	数据类型	货船	油船	准确率
VGG16Net	Z₃	0.732	0.768	0.739
GoogleNet	Z₃	0.722	0.736	0.727
ResNet18	Z₃	0.753	0.786	0.772
改进网络	Z₃	0.765	0.814	0.787

[1]	宋爽, 张悦, 张琳娜, 岑翼刚, 李浥东. 基于深度学习的轻量化目标检测算法[J]. 系统工程与电子技术, 2022, 44(9): 2716-2725.
[2]	苗添, 曾虹程, 王贺, 陈杰. 基于迭代阈值分割的星载SAR洪水区域快速提取[J]. 系统工程与电子技术, 2022, 44(9): 2760-2768.
[3]	聂倩, 杨丽花, 呼博, 任露露. 基扩展模型下基于LSTM神经网络的时变信道预测方法[J]. 系统工程与电子技术, 2022, 44(9): 2971-2977.
[4]	王健, 何自豪, 刘洁, 杨珂. 基于梯度域导向滤波器和改进PCNN的图像融合算法[J]. 系统工程与电子技术, 2022, 44(8): 2381-2392.
[5]	王彩云, 吴钇达, 王佳宁, 马璐, 赵焕玥. 基于改进的CNN和数据增强的SAR目标识别[J]. 系统工程与电子技术, 2022, 44(8): 2483-2487.
[6]	王冠, 茹海忠, 张大力, 马广程, 夏红伟. 弹性高超声速飞行器智能控制系统设计[J]. 系统工程与电子技术, 2022, 44(7): 2276-2285.
[7]	樊成, 王布宏, 田继伟. 基于多任务学习图卷积模型的航空网络节点分类[J]. 系统工程与电子技术, 2022, 44(7): 2341-2349.
[8]	金国栋, 薛远亮, 谭力宁, 许剑锟. 基于孪生神经网络的目标跟踪算法进展研究[J]. 系统工程与电子技术, 2022, 44(6): 1805-1822.
[9]	傅东宁, 廖桂生, 黄岩, 张邦杰, 王幸. 基于图拉普拉斯嵌入的合成孔径雷达时变窄带干扰抑制算法[J]. 系统工程与电子技术, 2022, 44(6): 1846-1853.
[10]	盖明慧, 张苏, 孙卫天, 倪育德, 杨磊. 复数兼容全变分SAR目标结构特征增强[J]. 系统工程与电子技术, 2022, 44(6): 1862-1872.
[11]	徐安林, 张毓, 周峰. 基于Beta过程的高分辨ISAR成像[J]. 系统工程与电子技术, 2022, 44(6): 1873-1879.
[12]	韦娟, 杨皇卫, 宁方立. 基于NMF与CNN联合优化的声学场景分类[J]. 系统工程与电子技术, 2022, 44(5): 1433-1438.
[13]	纪朋徽, 代大海, 邢世其, 冯德军. 密集虚假运动目标生成方法[J]. 系统工程与电子技术, 2022, 44(5): 1502-1511.
[14]	刘丰恺, 黄大荣, 郭新荣, 冯存前. 基于吕氏分布的机动目标参数化平动补偿方法[J]. 系统工程与电子技术, 2022, 44(4): 1166-1173.
[15]	陈冬, 句彦伟. 基于语义分割实现的SAR图像舰船目标检测[J]. 系统工程与电子技术, 2022, 44(4): 1195-1201.