浏览全部资源
扫码关注微信
收稿：2020-10-22，
修回：2020-12-25，
录用：2021-1-3，
纸质出版：2021-12-16
移动端阅览
目的
2
海洋锋的高效检测对海洋生态环境变化、渔业资源评估、渔情预报及台风路径预测等具有重要意义。海洋锋具有边界信息不明显且多变的弱边缘性，传统基于梯度阈值法及边缘检测的海洋锋检测方法，存在阈值选择不固定、判定指标不一致导致检测精度较低的问题。针对上述问题，基于Mask R-CNN（region convolutional neural network）提出一种改进的海洋锋自动检测方法。
方法
2
兼顾考虑海洋锋的小数据量及弱边缘性，首先对数据扩增，并基于不同算法对海表温度（sea surface temperatures，SST）遥感影像进行增强；其次，基于迁移学习的思想采用COCO（common objects in context）数据集对网络模型进行初始化；同时，对Mask R-CNN中残差神经网络（residual neural network，ResNet）和特征金字塔模型（feature pyramid network，FPN）分别进行改进，在充分利用低层特征高分辨率和高层特征的高语义信息的基础上，对多个尺度的融合特征图分别进行目标预测，提升海洋锋的检测精度。
结果
2
为验证本文方法的有效性，从训练数据和实验模型上分别设计多组对比实验。实验结果表明，相比常用的Mask R-CNN和YOLOv3（you only look once）神经网络，本文方法对SST梯度影像数据集上的海洋锋检测效果最好，海洋锋的定位准确率（intersection over union，IoU）及检测平均精度均值（mean average precision，mAP）达0.85以上。此外，通过对比分析实验结果发现，本文方法对强海洋锋的检测效果明显优于弱海洋锋。
结论
2
本文根据专家经验设立合理的海洋锋检测标准，更好地考虑了海洋锋的弱边缘性。通过设计多组对比实验，验证了本文方法对海洋锋的高精度检测效果。
Objective
2
The efficient detection of ocean front is of great significance to study the efficient detection of ocean front for marine ecosystem
fishery resources assessment
fishery forecast and typhoon track prediction. Gradient threshold method and edge detection algorithm have been widely used in ocean front detection. Traditional gradient method mainly depends on the gradient threshold
the sea area with gradient value greater than the set threshold has been regarded as the existence of ocean front. However
the selection criteria of threshold cannot tailor the requirements of accurate detection of complex and diverse ocean fronts due to artificial setting dependence
so it is more suitable for the object detection with fixed edge (such as land). During to the weak edge information of ocean front
it is difficult to achieve the good effect through the traditional edge extraction algorithm. A new automatic detection method to detect the small data volume and weak marginal characteristics of ocean fronts has been considering. Based on the advantages of the Mask R-CNN (region convolutional neural network) for instance segmentation
an improved Mask R-CNN network has been applied to the detection of ocean fronts. The ocean front detection method based on the modified Mask R-CNN has evolved the establishment of ocean front detection standards and data preprocessing
such as data expansion
data enhancement and labeling operations. High-precision detection of ocean fronts has been realized based on multiple iterations of training and parameter correction.
Method
2
First
the remote sensing images have been performed expansion operations for the small amount of data and the weak edge characteristics
such as rotating
flipping and cropping. Total 2 100 images have been obtained including 800 original images
500 rotation and flip processing images and 800 random cropping processing images. Meanwhile
sea surface temperatures (SST) remote sensing images have been enhanced based on deep closest point (DCP) and contrast limited adaptive histogram equalization (CLAHE) algorithms. Next
based on migration learning
using the general image classification network model trained on the common objects in context (COCO) dataset as the pre-training model
and using training datasets to train the pre-trained model. In order to meet the needs of ocean front detection
the residual network (ResNet) and feature pyramid network (FPN) model in Mask R-CNN have been optimized respectively. Limited training data leads to over fitting of the deep residual network and poor detection results. Considering the scarcity of ocean front data and the difficulty of constructing training set
the shallow ResNet-18 network has been used to detect ocean front. Multi-scale fusion feature maps have been predicted to enhance the detection effect of ocean front respectively through the full use of the high-resolution and high-level semantic information of low-level features.
Result
2
In order to verify the effectiveness of the method
three training datasets of grayscale image
RGB image and gradient image have been designed. Using LabelMe software to label the dataset
and then achieving high-precision detection of ocean fronts by multiple iterations of training and parameter correction. In addition
the weighted harmonic mean Micro-F1 and intersection over union (IoU) have both been used to evaluate the detection accuracy and target location accuracy of the model. In the experiment and analysis section
several groups of comparative experiments have been designed from experimental model. In order to evaluate the robustness and effectiveness of the method
images collection of global ocean fronts has been conducted to make three different datasets and perform multiple different iterations on the datasets. The results have shown that the training has effectively converged
and the detection accuracy of the model has also increased
reaching more than 0.85 after 25 000 iterations. In order to further verify the proposed ocean front detection results based on Mask R-CNN
the three training sets have been trained separately. The experimental results have shown that gradient images have been both higher than RGB images and grayscale images to detect ocean fronts
the positioning accuracy and detection accuracy. In order to highlight the advantages of this model for ocean front detection
this research has compared it with you only look once (YOLOv3) and Mask R-CNN models under three different datasets. The three training datasets have been trained for 30 000 times under different models. The results have demonstrated that the positioning accuracy IoU and accuracy F1 of the ocean front detection method proposed are improved. The detection accuracy of this method is 84.33%
and Micro-F1 is 86.57%. Compared with the YOLOv3 and Mask R-CNN algorithms
the Micro-F1 value has increased by 4.27% and 3.01% respectively. Rapid identification of ocean fronts is the key to practical fishery applications. The running time of the RGB image set under different network models and different iteration times has been presented. Under different iterations
the proposed model takes much less time than YOLOv3. At last
in order to evaluate the effectiveness of the method in the detection of strong and weak fronts
the strong and weak fronts in the three datasets have been screened and trained separately. The results have shown that the accuracy of the method in the detection of strong ocean fronts can be achieved all above 80% higher.
Conclusion
2
A reasonable ocean front detection standard has been setup combined with the weak edge characteristics of ocean front. By designing some comparative experiments to verify the high-precision detection effect of the method proposed in this paper on ocean fronts.
Bost C A, Cotté C, Bailleul F, Cherel Y, Charrassin J B, Guinet C, Ainley D G and Weimerskirch H. 2009. The importance of oceanographic fronts to marine birds and mammals of the southern oceans. Journal of Marine Systems, 78(3): 363-376[DOI:10.1016/j.jmarsys.2008.11.022]
Cao W D, Xie C, Han B and Dong J Y. 2020. Automatic fine recognition of ocean front fused with deep learning. Computer Engineering, 46(10): 266-274
曹维东, 解翠, 韩冰, 董军宇. 2020. 融合深度学习的自动化海洋锋精细识别. 计算机工程, 46(10): 266-274
Cayula J F and Cornillon P. 1995. Multi-image edge detections for SST images. Journal of Atmospheric and Oceanic Technology, 12(4): 821-829[DOI:10.1175/1520-0426(1995)012<0821:miedfs>2.0.co;2]
Chang P and Yan P F. 1996. Wavelet scale space filtering on edge detection. Pattern Recognition and Artificial Intelligence, 9(3): 251-257
常鹏, 阎平凡. 1996. 一种基于小波变换的多尺度边缘检测方法. 模式识别与人工智能, 9(3): 251-257
Davis L S. 1975. A survey of edge detection techniques. Computer Graphics and Image Processing, 4(3): 248-270[DOI:10.1016/0146-664X(75)90012-X]
Deng L, Wang Y Q, Liu Y, Wang F, Li S K and Liu J. 2019. A CNN-based vortex identification method. Journal of Visualization, 22(1): 65-78[DOI:10.1007/s12650-018-0523-1]
He K, Gkioxari G, Dollár P and Girshick R. 2017. Mask R-CNN//Proceedings of 2017 IEEE International Conference on Computer Vision (ICCV). Venice, Italy: IEEE: 2980-2988[ DOI: 10.1109/ICCV.2017.322 http://dx.doi.org/10.1109/ICCV.2017.322 ]
Hu Y, Shan Z L and Gao F. 2018. Ship detection basedon faster-RCNN and multiresolution SAR. Radio Engineering, 48(2): 96-100
胡炎, 单子立, 高峰. 2018. 基于Faster-RCNN和多分辨率SAR的海上舰船目标检测. 无线电工程, 48(2): 96-100)[DOI:10.3969/j.issn.1003-3106.2018.02.04]
Kostianoy A G, Ginzburg A I, Frankignoulle M and Delille B. 2004. Fronts in the Southern Indian Ocean as inferred from satellite sea surface temperature data. Journal of Marine Systems, 45(1/2): 55-73[DOI:10.1016/j.jmarsys.2003.09.004]
Lima E, Sun X, Dong J Y, Wang H, Yang Y T and Liu L P. 2017. Learning and transferring convolutional neural network knowledge to ocean front recognition. IEEE Geoscience and Remote Sensing Letters, 14(3): 354-358[DOI:10.1109/LGRS.2016.2643000]
Li A Z, Zhou W F and Fan X M. 2017. Research progress of methods for the extraction of mesoscale ocean fronts and eddies based on remote sensing data. Chinese Journal of Image Graphics, 22(6): 709-718
黎安舟, 周为峰, 范秀梅. 2017. 遥感图像中尺度海洋锋及涡旋提取方法研究进展. 中国图象图形学报, 22(6): 709-718)[DOI:10.11834/jig.160637]
Li Q Y, Zhong G Q and Xie C. 2019. Weak edge identification nets for ocean front detection[EB/OL]. [2020-09-30] . https://arxiv.org/pdf/1909.07827.pdf https://arxiv.org/pdf/1909.07827.pdf
Lima E, Sun X, Yang Y T and Dong J Y. 2017. Application of deep convolutional neural networks for ocean front recognition. Journal of Applied Remote Sensing, 11(4): #042610[DOI:10.1117/1.JRS.11.042610]
Liu Z. 2012. The Spatio-Temporal Variability of Oceanic Fronts offshore China Seas and Analysis of Marine Observations. Qingdao: Institute of Oceanography, Chinese Academy of Sciences
刘泽. 2012. 中国近海锋面时空特征研究及现场观测分析. 青岛: 中国科学院研究生院(海洋研究所)
Nieto K, Demarcq H and McClatchie S. 2012. Mesoscale frontal structures in the Canary Upwelling System: new front and filament detection algorithms applied to spatial and temporal patterns. Remote Sensing of Environment, 123: 339-346[DOI:10.1016/j.rse.2012.03.028]
Ou P, Lu K, Zhang Z and Liu Z Y. 2019. Target recognition and spatial location based on mask RCNN. Computer Measurement and Control, 27(6): 172-176
欧攀, 路奎, 张正, 刘泽阳. 2019. 基于Mask RCNN的目标识别与空间定位. 计算机测量与控制, 27(6): 172-176)[DOI:10.16526/j.cnki.11-4762/tp.2019.06.037]
Pi Q L and Hu J Y. 2010. Analysis of sea surface temperature fronts in the Taiwan Strait and its adjacent area using an advanced edge detection method. Science China Earth Sciences, 53(7): 1008-1016[DOI:10.1007/s11430-010-3060-x]
Ping B. 2015. Research on Oceanic Field Recovery and Frontal Detection. Wuhan: Wuhan University
平博. 2015. 海洋场恢复与锋面检测方法研究. 武汉: 武汉大学
Ping B, Su F Z, Meng Y S, Fang S H and Du Y Y. 2014. A model of sea surface temperature front detection based on a threshold interval. Acta Oceanologica Sinica, 33(7): 65-71[DOI:10.1007/s13131-014-0502-x]
Ren S H, Liu N and Wang H. 2015. Review of ocean front in Chinese marginal seas and frontal forecasting. Advances in Earth Science, 30(5): 552-563
任诗鹤, 刘娜, 王辉. 2015. 中国近海海洋锋和锋面预报研究进展. 地球科学进展, 30(5): 552-563)[DOI:10.11867/j.issn.1001-8166.2015.05.0552]
Shao L J, Zhang H L, Zhang C H and Zhou X. 2015. A method for detecting the oceanic front using remotely sensed sea-surface temperature. Hydrographic Surveying and Charting, 35(2): 42-44, 51
邵连军, 张红雷, 张春华, 周旋. 2015. 基于海温遥感资料的海洋锋检测方法. 海洋测绘, 35(2): 42-44, 51)[DOI:10.3969/j.issn.1671-3044.2015.02.011]
Sun X, Wang C G, Dong J Y, Lima E and Yang Y T. 2019. A multiscale deep framework for ocean fronts detection and fine-grained location. IEEE Geoscience and Remote Sensing Letters, 16(2): 178-182[DOI:10.1109/LGRS.2018.2869647]
Wang Z and He S X. 2004. An adaptive edge-detection method based on Canny algorithm. Journal of Image and Graphics, 9(8): 957-962
王植, 贺赛先. 2004. 一种基于Canny理论的自适应边缘检测方法. 中国图象图形学报, 9(8): 957-962)[DOI:10.3969/j.issn.1006-8961.2004.08.011]
Xue C J, Su F Z and Zhou J Q. 2007. Extraction of ocean fronts based on wavelet analysis. Marine Science Bulletin, 26(2): 20-27
薛存金, 苏奋振, 周军其. 2007. 基于小波分析的海洋锋形态特征提取. 海洋通报, 26(2): 20-27)[DOI:10.3969/j.issn.1001-6392.2007.02.003]
Zhang L Y. 2018. Characteristics and Influences of the Northwest Pacific Subtropical Sea Surface Temperature Front. Nanjing: Nanjing University of Information Science and Technology
张乐英. 2018. 西北太平洋副热带海洋锋的变化特征及其影响. 南京: 南京信息工程大学
Zhang W, Cao Y and Luo Y. 2014. An ocean front detection method based on the Canny operator and mathematical morphology. Marine Science Bulletin, 33(2): 199-203
张伟, 曹洋, 罗玉. 2014. 一种基于Canny和数学形态学的海洋锋检测方法. 海洋通报, 33(2): 199-203)[DOI:10.11840/j.issn.1001-6392.2014.02.013]
Zhao H. 2010. Research on Image Edge Detection Based on Mathematical Morphology. Dalian: Dalian University of Technology
赵慧. 2010. 基于数学形态学的图像边缘检测方法研究. 大连: 大连理工大学
Zhou L, Chen D K, Lei X T, Wang W, Wang G H and Han G J. 2019. Progress and perspective on interactions between ocean and typhoon. Chinese Science Bulletin, 64(1): 60-72
周磊, 陈大可, 雷小途, 王伟, 王桂华, 韩桂军. 2019. 海洋与台风相互作用研究进展. 科学通报, 64(1): 60-72
相关作者
相关机构
京公网安备11010802024621
