海洋科学中尺度涡的计算机视觉检测和分析方法

doi:10.11871/jfdc.issn.2096-742X.2020.06.004

数据与计算发展前沿 ›› 2020, Vol. 2 ›› Issue (6): 30-41.

doi: 10.11871/jfdc.issn.2096-742X.2020.06.004

• 专题:高性能计算在行业领域的特色应用 • 上一篇下一篇

海洋科学中尺度涡的计算机视觉检测和分析方法

沈飙(),陈扬,杨琛,刘博文

中国海洋大学,山东青岛 266100

收稿日期:2020-08-08 出版日期:2020-12-20 发布日期:2020-12-29
通讯作者: 沈飙
作者简介:沈飙,中国海洋大学,物理海洋教育部重点实验室计算中心,工程师,从事高性能计算方面的管理、维护与应用研究,具有丰富的海洋数值模式的编译、运行、应用支持等经验。本文负责总体统稿,提供作业平台和算法思路。
SHEN Biao is an engineer of Key Laboratory of physical Oceanography, MOE, China. He has been engaged in the management, maintenance and application research of HPC with rich experience in compiling, running and supporting Marine numerical models.In this paper, he is responsible for the overall draft, providing the operation platform and algorithm ideas.E-mail:shenbiao@ouc.edu.cn|陈扬,中国海洋大学信息科学与工程学院计算机科学与技术专业在读学士。本文负责在集群上开展实验,撰写初稿。
CHEN Yang is an undergraduate student majored in computer science and technology, School of information science and engineering, Ocean University of China. In this paper, he completed the first draft and parts of experiments and analysis.E-mail: chenyang8484@ouc.edu.cn|杨琛,中国海洋大学海洋与大气学院,硕士研究生。本文负责针对PAC竞赛调整实验,挖掘摘要引言的内容。
YANG Chen is a master student majored in marine science in the school of ocean and atmosphere, Ocean University of China. In this paper, he is responsible for the experimental adjustment for the PAC competition and composition of the abstract and introduction.E-mail: qdyangchen@163.com|刘博文,中国海洋大学海洋与大气学院,硕士研究生。本文负责后期修正文稿,整理实验结果,分析结论。
LIU Bowen is a master student majored in marine science from the school of ocean and atmosphere, Ocean University of China.In this paper, he is responsible for revising the manuscript, sorting out the experimental results and analyzing the conclusions.E-mail: 15610009099@163.com
基金资助:
自然资源部海洋灾害预报技术重点实验室开放基金项目“深度Gabor散射网络及其在海洋中尺度涡检测中的应用”(LOMF 1802)

Computer Vision Detection and Analysis of Mesoscale Eddies in Marine Science

SHEN Biao(),CHEN Yang,YANG Chen,LIU Bowen

Ocean University of China, Qingdao, Shandong 266100, China

Received:2020-08-08 Online:2020-12-20 Published:2020-12-29
Contact: SHEN Biao

摘要/Abstract

摘要：

【目的】中尺度涡是海洋科学领域一个重要的研究课题,在人工智能领域中深度神经网络的技术支持下,将计算机视觉与海洋领域相结合,促进中尺度涡智能识别技术的应用和发展。【方法】基于Mask-RCNN算法,将深度学习中的目标检测算法应用于中尺度涡检测,结合多模态卫星遥感图像数据,对海洋中的中尺度涡进行识别、分类和分割。【结果】传统的中尺度涡检测方法只能检测中尺度涡的位置和大小,本文算法利用多模态信息进行定位、分类和实例分割,有效提取中尺度涡的特征,可以获得更高的检测准确率。【结论】人工智能在中尺度涡识别和追踪问题上具有重大的应用潜力,但涡旋检测的评价指标仍需进一步完善,促进深度学习检测中尺度涡技术向快速高效的方向发展。

关键词: 中尺度涡, 目标检测算法, 深度学习, 多模态数据融合

Abstract:

[Objective] Mesoscale eddy is an important research topic in the field of marine science. With the support of deep neural network in the artificial intelligence field, computer vision is combined with ocean field technology to promote the application and development of intelligent recognition technology of mesoscale eddy. [Methods] Based on Mask-RCNN algorithm, a deep learning object detection algorithm is applied to the detection of mesoscale eddies. Combined with the multi-modal satellite remote sensing image data, the mesoscale eddies in the ocean can be identified, classified and segmented. [Results] The traditional method can only detect the location and size of mesoscale eddies. In this paper, multi-modal information is adopted to locate, classify and segment the mesoscale eddies. [Conclusions] Artificial intelligence has great application potential in recognizing and tracking mesoscale eddies. But the evaluation indexes of eddy detection still need to be further improved to promote the rapid and efficient development of deep learning detection of mesoscale eddies.

Key words: mesoscale eddy, object detection, deep learning, multimodal data fusion

沈飙,陈扬,杨琛,刘博文. 海洋科学中尺度涡的计算机视觉检测和分析方法[J]. 数据与计算发展前沿, 2020, 2(6): 30-41.

SHEN Biao,CHEN Yang,YANG Chen,LIU Bowen. Computer Vision Detection and Analysis of Mesoscale Eddies in Marine Science[J]. Frontiers of Data and Computing, 2020, 2(6): 30-41.

图/表 13

图1

表1

图2

图3

图4

图5

图6

图7

表2

图8

图9

表4

参考文献 32

[1]	王桂华, 苏纪兰, 齐义泉 . 南海中尺度涡研究进展[J]. 地球科学进展, 2005(08):882-886.
[2]	李立 . 南海中尺度海洋现象研究概述[J]. 台湾海峡, 2002(02):265-274.
[3]	郝滢洁 . 基于卷积神经网络的海洋中尺度涡旋检测算法研究[D]. 山东科技大学, 2017.
[4]	刘缵然 . 海洋遥感图像中尺度涡检测系统的设计与实现[D]. 青岛科技大学, 2018.
[5]	张慧 . 深度学习中优化算法的研究与改进[D]. 北京邮电大学, 2018.
[6]	Nicholas King . Impact investors seeing positive social, financial ROI[J]. Nonprofit Business Advisor, 2019(359):8.
[7]	Chao Yan, Weihai Chen, Peter C. Y. Chen, Kendrick Amezquita S., Xingming Wu,. A New Two-Stage Object Detection Network without RoI-Pooling[C]. 第30届中国控制与决策会议论文集( 2), 2018: 1680-1685.
[8]	Yuanyuan Qin. RoI Pooling Based Fast Multi-Domain Convolutional Neural Networks for Visual Tracking[A]. Science and Engineering Research Center.Proceedings of 2016 2nd International Conference on Artificial Intelligence and Industrial Engineering (AIIE2016) [C].Science and Engineering Research Center:Science and Engineering Research Center, 2016: 5.
[9]	Yuanyuan Qin, Shiying He, Yong Zhao, Yuanzhi Gong. RoI Pooling Based Fast Multi-Domain Convolutional Neural Networks for Visual Tracking[P]. Proceedings of the 2016 2nd International Conference on Artificial Intelligence and Industrial Engineering (AIIE 2016), 2016.
[10]	Muhammad Anwar, Ms. Ramdani, Ahmad Fudhail Majid. The Effect of using Interactive Students’ Worksheet on Learning Achievement and Students’ Perception in The Basic Chemistry Course (Study on Reaction Rate)[P]. Proceedings of the 1st International Conference on Advanced Multidisciplinary Research (ICAMR 2018), 2018.
[11]	Yi Hou, Hong Zhang, Shilin Zhou, Huanxin Zou, Paolo Bellavista . Efficient ConvNet Feature Extraction with Multiple RoI Pooling for Landmark-Based Visual Localization of Autonomous Vehicles[J]. Mobile Information Systems, 2017: 1-14.
[12]	陈怡佳 . 基于Faster RCNN的目标检测系统[D]. 哈尔滨理工大学, 2019.
[13]	郑书新 . 针对深度学习模型的优化问题研究[D]. 中国科学技术大学, 2019.
[14]	丁鹏 . 基于深度卷积神经网络的光学遥感目标检测技术研究[D]. 中国科学院大学(中国科学院长春光学精密机械与物理研究所), 2019.
[15]	Xudong Sun, Pengcheng Wu, Steven C. H. Hoi. Face detection using deep learning: An improved faster RCNN approach[J]. Neurocomputing, 2018(299):42-50.
[16]	Sotiria Georgiou, Carine G. van der Boog, Nils Brüggemann, Stefanie L. Ypma, Julie D,. Pietrzak,Caroline A. Katsman. On the interplay between downwelling, deep convection and mesoscale eddies in the Labrador Sea[J]. Ocean Modelling, 2019(135):56-70.
[17]	Frantisek Jabloncik, Libor Hargas, Dusan Koniar, Jozef Volak . Detection of static objects in an image using texture analysis[J]. Transportation Research Procedia, 2019 ( 40):265-270.
[18]	Liqin Yang, Nong Sang, Changxin Gao . Vehicle parts detection based on Faster - RCNN with location constraints of vehicle parts feature point[P]. International Symposium on Multispectral Image Processing and Pattern Recognition, 2018.
[19]	Xiyang Zhu. Server monitoring system using an improved Faster RCNN approach[A]. Xiamen Institute of Electronics, Institute of Modern Circuit & System Technology, Xiamen University.Proceedings of 2017 11th IEEE International Conference on Anti-counterfeiting, Security,and Identification(ASID)[C]. Xiamen Institute of Electronics, Institute of Modern Circuit & System Technology, Xiamen University:IEEE BEIJING SECTION(跨国电气电子工程师学会北京分会), 2017: 4.
[20]	Hong Xiao. Identification of Dangerous Goods in Human THZ Images[A]. Wuhan Zhicheng Times Cultural Development Co., Ltd.Proceedings of 2018 International Conference on Network, Communication, Computer Engineering (NCCE 2018)[C]. Wuhan Zhicheng Times Cultural Development Co., Ltd:武汉志诚时代文化发展有限公司, 2018: 5.
[21]	Faghmous J H, Styles L, Mithal V, et al. EddyScan: A physically consistent ocean eddy monitoring application[C]. 2012 Conference on Intelligent Data Understanding IEEE, 2012: 96-103.
[22]	Anderson S.P., Cooper C.K., Driver D.B., van Smirren, J. EddyNet: a measurement program for collecting real-time current profiles from the Gulf of Mexico[P]. Current Measurement Technology, 2003. Proceedings of the IEEE/OES Seventh Working Conference on, 2003.
[23]	Faghmous J.H., Styles L., Mithal V., Boriah S., Liess S., Kumar V., Vikebo F., dos Santos Mesquita, M. EddyScan: A physically consistent ocean eddy monitoring application[P]. Intelligent Data Understanding (CIDU), 2012 Conference on, 2012.
[24]	Qidan Zhu, Yanke Wang, Yunqian He, Xiao Hong . Object tracking with particles weighted by region proposal network[J]. Multimedia Tools and Applications , 2019(78):12083-12101.
[25]	Hedieh Sajedi, Fatemeh Mohammadipanah , Seyyed Amir Hosein Rahimi. Actinobacterial strains recognition by Machine learning methods[J]. Multimedia Tools and Applications, 2019(78):20285-20307.
[26]	He K, Zhang X, Ren S, et al. Deep residual learning for image recognition[C]// Proceedings of the IEEE conference on computer vision and pattern recognition. 2016: 770-778.
[27]	Ri-Gui Zhou, Yu Cheng , DaQian Liu. Quantum image scaling based on sbilinear interpolation with arbitrary scaling ratio[J]. Quantum Information Processing, 2019(18):267-285.
[28]	Ren S, He K, Girshick R , et al. Faster R-CNN: towards real-time object detection with region proposal networks[C]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017(39):1137-1149.
[29]	Kim K.-H., Shim P.-S., Shin S. An Alternative Bilinear Interpolation Method Between Spherical Grids[J]. Atmosphere 2019( 10):123-132.
[30]	Wei Qifen . Parametric Design and Application of Steel Anchor Box for Main Girder of Long Span Cable-Stayed Bridge[J]. IOP Conference Series: Materials Science and Engineering, 2018,392:062007. doi: 10.1088/1757-899X/392/6/062007
[31]	He K, Gkioxari, G, Dollar P, et al. Mask R-CNN[J]. Proceedings of the IEEE International Conference on Computer Vision (ICCV), 2017(99):2961-2969.
[32]	Yanling Du, Wei Song, Qi He, Dongmei Huang, Antonio Liotta, Chen Su , Deep learning with multi-scale feature fusion in remote sensing for automatic oceanic eddy detection[J]. Information Fusion, 2019(49):89-99.

锚点规模	学习率（头网络）	学习率（整体训练）	mAP （%）
（8,32,128）	0.001	0.001	45.50
（4,16,64）	0.001	0.001	53.89
（8,32,128）	0.001	0.0001	52.62
（4,16,64）	0.001	0.0001	56.35

α、β、γ比例	mAP	α、β、γ比例	mAP
1:1:1	60.48	3:5:2	60.15
2:2:1	58.69	2:3:5	57.75
2:1:2	58.70	3:2:5	55.76
1:2:2	57.43	1:1:0	57.43
2:1:1	66.85	1:0:1	56.48
1:2:1	59.62	0:1:1	52.91
1:1:2	58.62	1:0:0	56.32
5:3:2	70.27	0:1:0	44.91
5:2:3	67.47	0:0:1	49.26
2:5:3	59.73

	DeepEddy	EddyNet	本方法
涡旋分类	√	√	√
位置预测	×	×	√
涡旋分割	×	√	√
图像处理	√	√	√
遥感数据处理	×	×	√

海洋科学中尺度涡的计算机视觉检测和分析方法

Computer Vision Detection and Analysis of Mesoscale Eddies in Marine Science

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 13

参考文献 32

相关文章 15

编辑推荐

Metrics

本文评价

[1]	许淞源,刘峰. ESDRec：一种面向地球大数据平台的数据推荐模型[J]. 数据与计算发展前沿, 2023, 5(1): 55-64.
[2]	陈琼,杨咏,黄天林,冯媛. 小样本图像语义分割综述[J]. 数据与计算发展前沿, 2021, 3(6): 17-34.
[3]	蒲晓蓉,黄佳欣,刘军池,孙家瑜,罗纪翔,赵越,陈柯成,任亚洲. 面向临床需求的CT图像降噪综述[J]. 数据与计算发展前沿, 2021, 3(6): 35-49.
[4]	何涛,王桂芳,马廷灿. 基于词嵌入语义异常的跨学科研究内容发现方法[J]. 数据与计算发展前沿, 2021, 3(6): 50-59.
[5]	张怡宁,何洪波,王闰强. 热门数字音频预测技术综述[J]. 数据与计算发展前沿, 2021, 3(4): 81-92.
[6]	陈子健,李俊,岳兆娟,赵泽方. 基于自编码器与属性信息的混合推荐模型[J]. 数据与计算发展前沿, 2021, 3(3): 148-155.
[7]	肖建平,龙春,赵静,魏金侠,胡安磊,杜冠瑶. 基于深度学习的网络入侵检测研究综述[J]. 数据与计算发展前沿, 2021, 3(3): 59-74.
[8]	李序,连一峰,张海霞,黄克振. 网络安全知识图谱关键技术[J]. 数据与计算发展前沿, 2021, 3(3): 9-18.
[9]	赵伟昱,张宏海,仲波. 基于深度学习的遥感影像地块分割方法[J]. 数据与计算发展前沿, 2021, 3(2): 133-141.
[10]	任荟颖,王婧,王彦棡. 基于AutoML的湍流建模[J]. 数据与计算发展前沿, 2020, 2(4): 121-131.
[11]	张圣林,林潇霏,孙永谦,张玉志,裴丹. 基于深度学习的无监督KPI异常检测[J]. 数据与计算发展前沿, 2020, 2(3): 87-100.
[12]	陈雷,袁媛. 基于深度迁移学习的农业病害图像识别[J]. 数据与计算发展前沿, 2020, 2(2): 111-119.
[13]	刘成林. 文档图像识别技术回顾与展望[J]. 数据与计算发展前沿, 2019, 1(2): 17-25.
[14]	俞益洲, 马杰超, 石德君, 周振. 深度学习在医学影像分析中的应用综述[J]. 数据与计算发展前沿, 2019, 1(2): 37-52.
[15]	马艳军,于佃海,吴甜,王海峰. 飞桨：源于产业实践的开源深度学习平台[J]. 数据与计算发展前沿, 2019, 1(1): 105-115.