基于地基气辉图像的大气重力波目标识别

doi:10.19678/j.issn.1000-3428.0066521

摘要/Abstract

摘要：

子午工程全天空气辉成像仪观测网的逐步完善，积累了海量的气辉图像原始观测数据，但是基于地基气辉观测的大气重力波研究极度依赖人工识别，十分耗时且识别质量难以得到保证，亟需发展一种快速有效的自动识别方法。针对大气重力波标注样本稀缺的问题，在改进Cycle GAN模型的基础上提出一种大气重力波气辉观测数据集扩充算法，仅需标注少量样本即可大幅提升检测算法对大气重力波的识别精度；进一步，利用地基气辉图像识别目标与背景低信噪比的特点，通过对深度学习模型YOLOv5s骨干网络以及边界框预测加以改进，提出一种新的大气重力波智能识别算法。实验结果表明，使用扩增的数据集以及改进的YOLOv5s目标检测算法，在交并比阈值为0.5的情况下平均识别精度达到75.8%，较原始模型提升9.7个百分点，检测速度和平均识别精度均优于对比的主流目标检测算法。

关键词: 重力波, 深度学习, 数据扩充, 目标检测, 生成对抗网络

Abstract:

With the construction of the Meridian Project all-sky airglow imager observation network, a large amount of raw airglow image data has been accumulated. The current atmospheric gravity wave research based on airglow observation is extremely dependent on manual identification, which is very time-consuming, and the quality of labeling is difficult to guarantee. Therefore, there is an urgent need for a fast and effective automatic identification method. To solve the problem of sparsely labeled samples of atmospheric gravity waves, this paper proposes an algorithm based on the improved Cycle GAN model to expand the atmospheric gravity wave airglow observation dataset, thereby greatly improving the recognition accuracy of atmospheric gravity waves by labeling only a small number of samples. A new intelligent recognition algorithm for atmospheric gravity waves is also proposed by improving the YOLOv5s model backbone network and bounding box prediction, considering the characteristics of low Signal-to-Noise Ratio(SNR) between the recognition target and background in airglow images. The experimental results showed that using the augmented dataset and improved YOLOv5s target detection algorithm, the average precision reached 75.8% under an Intersection-over-Union(IoU) threshold of 0.5, which is 9.7 percentage points higher than that of the original model. Meanwhile, the detection speed and average recognition accuracy are superior to mainstream target detection algorithms compared.

Key words: gravity wave, deep learning, data augmentation, object detection, Generative Adversarial Network(GAN)

陈锦生, 马文臻, 方少峰, 邹自明. 基于地基气辉图像的大气重力波目标识别[J]. 计算机工程, 2023, 49(11): 13-23.

Jinsheng CHEN, Wenzhen MA, Shaofeng FANG, Ziming ZOU. Object Recognition of Atmospheric Gravity Wave Based on Foundation Airglow Images[J]. Computer Engineering, 2023, 49(11): 13-23.

http://www.ecice06.com/CN/Y2023/V49/I11/13

图/表 16

图1 Cycle GAN流程

Fig.1 Procedure of Cycle GAN

图2 本文算法框架

Fig.2 Framework of the proposed algorithm

图3 部分模块结构

Fig.3 Structures of some modules

图4 BiFPN结构

Fig.4 Structure of BiFPN

图5 WBF示意图

Fig.5 Schematic diagram of WBF

图6 实验流程

Fig.6 Procedure of experiment

图7 模型生成器G和判别器D_Y损失函数趋势

Fig.7 Loss function trend of model generator G and discriminator D_Y

图8 不同方法生成的样本图

Fig.8 Sample pictures generated by different methods

图9 预测框回归损失/目标检测损失/AP@0.5随训练迭代次数的变化

Fig.9 Prediction box regression loss/target detection loss/AP@0.5 variation with training iterations

图10 4种算法检测结果对比

Fig.10 Comparison of detection results of four algorithms

参考文献 26

1	NAPPO C J. An introduction to atmospheric gravity waves. Cambridge, USA: Academic Press, 2002.
2	FRITTS D C, ALEXANDER M J. Gravity wave dynamics and effects in the middle atmosphere. Reviews of Geophysics, 2003, 41 (1): 1- 64. doi: 10.1029/2001RG000106
3	夏重庆. 大气重力波对飞机纵向稳定性的影响[D]. 南京: 南京航空航天大学, 2012.
	XIA C Q. Influence of atmospheric gravity wave on longitudinal stability of aircraft[D]. Nanjing: Nanjing University of Aeronautics and Astronautics, 2012. (in Chinese)
4	殷青青. 中层大气重力波信号的提取和数值模拟研究[D]. 兰州: 兰州大学, 2022.
	YIN Q Q. Study on extraction and numerical simulation of gravity wave signal in the middle atmosphere[D]. Lanzhou: Lanzhou University, 2022. (in Chinese)
5	PETERSON A W, ADAMS G W. OH airglow phenomena during the 5-6 July 1982 total lunar eclipse. Applied Optics, 1983, 22 (17): 2682- 2685. doi: 10.1364/AO.22.002682
6	WANG C, CHEN Z Q, XU J Y. Recent advances in research of the Chinese meridian project. Chinese Journal of Space Science, 2020, 40 (5): 679- 690. doi: 10.11728/cjss2020.05.679
7	ALZUBAIDI L, ZHANG J L, HUMAIDI A J, et al. Review of deep learning: concepts, CNN architectures, challenges, applications, future directions. Journal of Big Data, 2021, 8 (1): 1- 74. doi: 10.1186/s40537-020-00387-6
8	SUN X F, YANG D Y, LI X Y, et al. Interpreting deep learning models in natural language processing: a review[EB/OL]. [2022-06-10]. https://arxiv.org/abs/2110.10470.
9	CHANG L, XU J Y, YUE J A, et al. Automatic extraction of gravity waves from all-sky airglow image based on machine learning. Remote Sensing, 2019, 11 (13): 1516. doi: 10.3390/rs11131516
10	HEATON J, GOODFELLOW I, BENGIO Y, et al. Deep learning. Cambridge, USA: MIT Press, 2016.
11	LIN T Y, MAIRE M, BELONGIE S, et al. Microsoft COCO: common objects in context[C]//Proceedings of European Conference on Computer Vision. Berlin, Germany: Springer, 2014: 740-755.
12	TAYLOR L, NITSCHKE G. Improving deep learning with generic data augmentation[C]//Proceedings of IEEE Symposium Series on Computational Intelligence. Washington D. C., USA: IEEE Press, 2019: 1542-1547.
13	GOODFELLOW I J, POUGET-ABADIE J, MIRZA M, et al. Generative adversarial nets[C]//Proceedings of the 27th International Conference on Neural Information Processing Systems. New York, USA: ACM Press, 2014: 2672-2680.
14	KAUR P, KHEHRA B S, MAVI E B S. Data augmentation for object detection: a review[C]//Proceedings of IEEE International Midwest Symposium on Circuits and Systems. Washington D. C., USA: IEEE Press, 2021: 537-543.
15	洪硕. 高分辨率可见光遥感图像船舶目标数据增强方法研究[D]. 哈尔滨: 哈尔滨工业大学, 2020.
	HONG S. Research on ship target data enhancement method of high resolution visible light remote sensing image[D]. Harbin: Harbin Institute of Technology, 2020. (in Chinese)
16	黄攀, 杨小冈, 卢瑞涛, 等. 基于GAN的红外飞机数据增强方法. 电光与控制, 2021, 28 (11): 84- 88. doi: 10.3969/j.issn.1671-637X.2021.11.018
	HUANG P, YANG X G, LU R T, et al. Data augmentation of infrared aircraft target based on generative adversarial network. Electronics Optics & Control, 2021, 28 (11): 84- 88. doi: 10.3969/j.issn.1671-637X.2021.11.018
17	ZHANG H B, PAN D, LIU J H, et al. A novel MAS-GAN-based data synthesis method for object surface defect detection. Neurocomputing, 2022, 499, 106- 114. doi: 10.1016/j.neucom.2022.05.021
18	DUAN B Q, PAN W L. A data preprocessing method and preliminary results of all-sky airglow image. Chinese Journal of Space Science, 2017, 37 (1): 94. doi: 10.11728/cjss2017.01.094
19	LI Q, XU J, YUE J, et al. Statistical characteristics of gravity wave activities observed by an OH airglow imager at Xinglong, in Northern China. Annales Geophysicae, 2011, 29 (8): 1401- 1410. doi: 10.5194/angeo-29-1401-2011
20	GUI J, SUN Z N, WEN Y G, et al. A review on generative adversarial networks: algorithms theory and applications. IEEE Transactions on Knowledge and Data Engineering, 2023, 35 (4): 3313- 3332. doi: 10.1109/TKDE.2021.3130191
21	ZHU J Y, PARK T, ISOLA P, et al. Unpaired image-to-image translation using cycle-consistent adversarial networks[C]//Proceedings of IEEE International Conference on Computer Vision. Washington D. C., USA: IEEE Press, 2017: 2242-2251.
22	MAO X D, LI Q, XIE H R, et al. Least Squares generative adversarial networks[C]//Proceedings of IEEE International Conference on Computer Vision. Washington D. C., USA: IEEE Press, 2017: 2813-2821.
23	REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once: unified, real-time object detection[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2016: 779-788.
24	GLENN J, ALEX S, JIRKA B, et al. ultralytics/yolov5[EB/OL]. (2020-05-18)[2022-12-05]. https://github.com/ultralytics/yolov5.
25	TAN M X, PANG R M, LE Q V. EfficientDet: scalable and efficient object detection[C]//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2020: 10778-10787.
26	SOLOVYEV R, WANG W M, GABRUSEVA T. Weighted boxes fusion: ensembling boxes from different object detection models. Image and Vision Computing, 2021, 107, 1- 3. doi: 10.48550/arXiv.1910.13302

[1]	李嘉新, 侯进, 盛博莹, 周宇航. 基于改进YOLOv5的遥感小目标检测网络[J]. 计算机工程, 2023, 49(9): 256-264.
[2]	池亚平, 岳梓岩, 林雨衡. 基于Transformer的SM4算法工作模式识别[J]. 计算机工程, 2023, 49(9): 109-117.
[3]	林中霖, 时金桥, 王美琪, 王学宾, 王雨燕. 基于应用行为划分的Android恶意应用检测技术[J]. 计算机工程, 2023, 49(9): 125-136.
[4]	龙玉江, 卫薇, 舒彧, 张正刚, 王道累, 李峰. 基于自适应关键点的破损旋转绝缘子检测方法[J]. 计算机工程, 2023, 49(9): 272-278.
[5]	徐春波, 闫娟, 杨慧斌, 王博, 吴晗. 基于目标检测和语义分割的视觉SLAM算法[J]. 计算机工程, 2023, 49(8): 199-206, 214.
[6]	江雨燕, 陶承凤, 李平. 数据增强和自适应自步学习的深度子空间聚类算法[J]. 计算机工程, 2023, 49(8): 96-103, 110.
[7]	宋志娜, 李莎, 杨建明, 徐川. 基于特征与区域定位增强的遥感舰船目标检测[J]. 计算机工程, 2023, 49(8): 257-264.
[8]	李泽水, 冀俊忠, 杨翠翠. 基于边权重信息深度网络嵌入的PPIN功能模块检测[J]. 计算机工程, 2023, 49(8): 69-76.
[9]	王可铮, 徐玉芬, 周尚波. 结合对比感知损失和融合注意力的图像去雾模型[J]. 计算机工程, 2023, 49(8): 207-214.
[10]	刘俊豪, 王美林, 谢兴, 宋烨兴, 许莉花. 基于改进YOLOv5的皮革瑕疵检测算法[J]. 计算机工程, 2023, 49(8): 240-249.
[11]	李强龙, 周新文, 位梦恩, 甘阳洲. 基于条形池化和注意力机制的街道场景红外目标检测算法[J]. 计算机工程, 2023, 49(8): 310-320.
[12]	闫兴亚, 匡娅茜, 白光睿, 李月. 基于深度学习的学生课堂行为识别方法[J]. 计算机工程, 2023, 49(7): 251-258.
[13]	聂志勇, 阴宇薇, 汤佳欣, 涂志刚. 一种基于边界框关键点距离的框回归算法[J]. 计算机工程, 2023, 49(7): 65-75.
[14]	李军侠, 王星驰, 殷梓, 石德硕. 边缘深度挖掘的弱监督显著性目标检测[J]. 计算机工程, 2023, 49(7): 169-178.
[15]	吴珊, 周凤. 基于改进SSD算法的小目标检测[J]. 计算机工程, 2023, 49(7): 179-188.

选择文件类型/文献管理软件名称

选择包含的内容