Research on Joint Object Detection Method for Multi-Band Images of Space Hurricane

doi:10.11871/jfdc.issn.2096-742X.2025.04.010

Abstract

Abstract:

[Objective] As a typical phenomenon triggered by solar-terrestrial interactions, space hurricane often generates a huge energy cyclone in the middle and upper atmosphere, which is accompanied by the occurrence of auroral phenomena. Identifying auroral images can assist scientists in finding typical space hurricane events. However, currently, the search for such events mainly relies on experts' manual identification of auroral images, which is rather inefficient. To solve the above problems, this study explores a deep learning-based joint object detection method for multi-band images, achieving event recognition and precise localization of space hurricanes. [Methods] In this study, four-band auroral images (121.6nm, 135.6nm, LBHS, LBHL) from DMSP/SSUSI are used to identify space hurricane events. Based on the YOLOv8 algorithm framework, target-level fusion and feature-level fusion strategies are introduced. Meanwhile, single-band and multi-band fusion models for space hurricane recognition are established. [Results] In the event recognition task, by comparing the experimental results of single-band baseline models and multi-band fusion models, it is shown that the 1216_LBHL combination in feature-level fusion performs best, with an F1 score of 0.941. In the object detection task, the 1216_LBHL combination in target-level fusion achieves the highest AP value of 0.917. [Conclusions] Feature-level fusion demonstrates greater advantages in space hurricane event recognition, while target-level fusion is more suitable for object detection tasks. This indicates that the combined optimization of multi-band complementarity and fusion strategies is the key to enhancing detection performance.

Key words: space hurricane, YOLOv8, target-level fusion, feature-level fusion

SHI Ke, LU Yang, LU Sheng, WANG Yong, ZOU Ziming. Research on Joint Object Detection Method for Multi-Band Images of Space Hurricane[J]. Frontiers of Data and Computing, 2025, 7(4): 118-128, https://cstr.cn/32002.14.jfdc.CN10-1649/TP.2025.04.010.

Figures/Tables 11

Fig.1

Fig.2

Fig.3

Fig.4

Fig.5

Fig.6

Table 1

Table 2

Table3

Table 4

Fig.7

References 21

[1]	孔婉秋. 极光辐射机理研究及多波段光谱数据应用[D]. 西安: 西安电子科技大学, 2020.
[2]	ZHANG Q H, ZHANG Y L, WANG C, et al. A space hurricane over the Earth’s polar ionosphere[J]. Nature Communications, 2021, 12(1): 1-10.
[3]	LU S, XING Z Y, ZHANG Q H, et al. A statistical study of space hurricanes in the Northern Hemisphere[J]. Frontiers in Astronomy and Space Sciences, 2022, 9: 1047982.
[4]	LECUN Y, BOTTOU L, BENGIO Y, et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE, 1998, 86(11): 2278-2324.
[5]	佟欣, 邹自明, 白曦, 等. 喉区极光的机器识别[J]. 空间科学学报, 2021, 41(4): 654-666.
[6]	任杰. 基于深度学习的极光事件自动识别与检测[D]. 西安: 陕西师范大学, 2022.
[7]	夏凯, 邢赞扬, 张清和, 等. 基于迁移学习的太空台风自动识别[J]. 空间科学学报, 2023, 43(2): 231-240.
[8]	REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once: Unified, real-time object detection[C]// Proceedings of the IEEE conference on computer vision and pattern recognition,2016: 779-788.
[9]	LIU W, ANGUELOV D, ERHAN D, et al. Ssd: Single shot multibox detector[C]// Computer Vision-ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, Springer International Publishing, 2016: 21-37.
[10]	REN S, HE K, GIRSHICK R, et al. Faster r-cnn: Towards real-time object detection with region proposal networks[J]. Advances in Neural Information Processing Systems, 2015, 28.
[11]	HE K, GKIOXARI G, DOLLÁR P, et al. Mask r-cnn[C]// Proceedings of the IEEE international conference on computer vision. 2017: 2961-2969.
[12]	LU Y, JIANG J, ZHONG J, et al. Automatic detection method of polar cap arc based on YOLOX embedded with CBAM[J]. Frontiers in Environmental Science, 2024, 12: 1418207.
[13]	VARGHESE R, SAMBATH M. YOLOv8: A Novel Object Detection Algorithm with Enhanced Performance and Robustness[C]//2024 International Conference on Advances in Data Engineering and Intelligent Computing Systems (ADICS). IEEE, 2024: 1-6.
[14]	RAMACHANDRAM D, TAYLOR G W. Deep multimodal learning: A survey on recent advances and trends[J]. IEEE Signal Processing Magazine, 2017, 34(6): 96-108.
[15]	PAXTON L J, MORRISON D, ZHANG Y, et al. Validation of remote sensing products produced by the Special Sensor Ultraviolet Scanning Imager (SSUSI): A far UV-imaging spectrograph on DMSP F-16[C]// Optical spectroscopic techniques, remote sensing, and instrumentation for atmospheric and space research IV. SPIE, 2002, 4485: 338-348.
[16]	TORRALBA A, RUSSELL B C, YUEN J. Labelme: Online image annotation and applications[J]. Proceedings of the IEEE, 2010, 98(8): 1467-1484.
[17]	WANG C Y, LIAO H Y M, WU Y H, et al. CSPNet: A new backbone that can enhance learning capability of CNN[C]// Proceedings of the IEEE/CVF conference on computer vision and pattern recognition workshops,2020: 390-391.
[18]	LIN T Y, DOLLÁR P, GIRSHICK R, et al. Feature pyramid networks for object detection[C]// Proceedings of the IEEE conference on computer vision and pattern recognition,2017: 2117-2125.
[19]	LIU S, QI L, QIN H, et al. Path aggregation network for instance segmentation[C]// Proceedings of the IEEE conference on computer vision and pattern recognition,2018: 8759-8768.
[20]	NEUBECK A, VAN GOOL L. Efficient non-maximum suppression[C]//18th international conference on pattern recognition (ICPR'06). IEEE, 2006, 3: 850-855.
[21]	母健宇, 赵博雅, 贺志鹏, 等. 多模态渐近融合轻量化遥感图像车辆目标检测[C]// 中国高科技产业化研究会智能信息处理产业化分会. 第十八届全国信号和智能信息处理与应用学术会议论文集, 2024: 521-527.

波段	Precision	Recall	F1值
1216	1.000	0.056	0.105
1356	0.857	0.833	0.845
LBHS	0.882	0.833	0.857
LBHL	0.903	0.778	0.836

波段组合	融合方法	Precision	Recall	F1值
1216_1356	目标级融合	1.000	0.086	0.158
1216_LBHS	目标级融合	0.861	0.886	0.873
1216_LBHL	目标级融合	0.868	0.943	0.904
1356_LBHS	目标级融合	0.775	0.886	0.827
1356_LBHL	目标级融合	0.825	0.943	0.880
LBHS_LBHL	目标级融合	0.906	0.829	0.866
1216_1356_LBHS	目标级融合	1.000	0.829	0.906
1216_1356_LBHL	目标级融合	1.000	0.800	0.889
1216_LBHS_LBHL	目标级融合	0.865	0.914	0.889
1356_LBHS_LBHL	目标级融合	0.806	0.829	0.817
1216_1356_LBHS_LBHL	目标级融合	0.941	0.914	0.928
1216_1356	特征级融合	1.000	0.829	0.906
1216_LBHS	特征级融合	0.872	0.971	0.919
1216_LBHL	特征级融合	0.970	0.914	0.941
1356_LBHS	特征级融合	0.968	0.857	0.909
1356_LBHL	特征级融合	0.846	0.943	0.892
LBHS_LBHL	特征级融合	0.857	0.857	0.857
1216_1356_LBHS	特征级融合	0.906	0.829	0.866
1216_1356_LBHL	特征级融合	0.912	0.886	0.899
1216_LBHS_LBHL	特征级融合	0.912	0.886	0.899
1356_LBHS_LBHL	特征级融合	0.939	0.886	0.912
1216_1356_LBHS_LBHL	特征级融合	0.886	0.886	0.886

波段	Precision	Recall	AP
1216	1.000	0.056	0.528
1356	0.800	0.778	0.828
LBHS	0.808	0.778	0.823
LBHL	0.839	0.722	0.798

波段组合	融合方法	Precision	Recall	AP
1216_1356	目标级融合	1.000	0.086	0.543
1216_LBHS	目标级融合	0.806	0.829	0.831
1216_LBHL	目标级融合	0.846	0.886	0.917
1356_LBHS	目标级融合	0.809	0.743	0.792
1356_LBHL	目标级融合	0.732	0.857	0.824
LBHS_LBHL	目标级融合	0.844	0.771	0.841
1216_1356_LBHS	目标级融合	0.931	0.771	0.849
1216_1356_LBHL	目标级融合	0.964	0.771	0.872
1216_LBHS_LBHL	目标级融合	0.837	0.800	0.828
1356_LBHS_LBHL	目标级融合	0.806	0.829	0.870
1216_1356_LBHS_LBHL	目标级融合	0.968	0.800	0.881
1216_1356	特征级融合	0.862	0.714	0.792
1216_LBHS	特征级融合	0.822	0.857	0.854
1216_LBHL	特征级融合	0.848	0.800	0.809
1356_LBHS	特征级融合	0.903	0.800	0.868
1356_LBHL	特征级融合	0.792	0.800	0.804
LBHS_LBHL	特征级融合	0.912	0.714	0.829
1216_1356_LBHS	特征级融合	0.909	0.771	0.868
1216_1356_LBHL	特征级融合	0.878	0.826	0.871
1216_LBHS_LBHL	特征级融合	0.878	0.826	0.871
1356_LBHS_LBHL	特征级融合	0.857	0.857	0.852
1216_1356_LBHS_LBHL	特征级融合	0.829	0.829	0.873