基于YOLACT的冰体环向裂纹尺寸识别方法

周利; 蔡金延; 丁仕风; 刘仁伟; 曾鼎翰

doi:10.19693/j.issn.1673-3185.03024

基于YOLACT的冰体环向裂纹尺寸识别方法

doi: 10.19693/j.issn.1673-3185.03024

1.
上海交通大学船舶海洋与建筑工程学院，上海 200240
2.
江苏科技大学船舶与海洋工程学院，江苏镇江 212100

基金项目: 国家重点研发计划资助项目（2022YFE010700）；国家自然科学基金面上资助项目（52171259）

详细信息

作者简介:
周利，男，1983 年生，博士，教授。研究方向：极地船舶与海洋工程。E-mail：zhouli209@hotmail.com

蔡金延，男，1998年生，硕士。研究方向：计算机视觉，海冰图像处理。E-mail：209010066@stu.just.edn.cn

丁仕风，男，1981 年生，博士，正高级工程师。研究方向：极地船舶与海洋工程。E-mail：15001945469@163.com

通信作者:
丁仕风

中图分类号: U666.157
计量
- 文章访问数: 164
- HTML全文浏览量: 17
- PDF下载量: 3
- 被引次数: 0
出版历程
- 收稿日期: 2022-07-31
- 修回日期: 2022-09-06
- 网络出版日期: 2022-09-07

Method of recognizing ice circumferential crack size based on YOLACT

1.
School of Marine and Architecture Engineering, Shanghai Jiaotong University, Shanghai 200240, China
2.
School of Naval Architecture and Ocean Engineering, Jiangsu University of Science and Technology, Zhenjiang 212100, China

基于YOLACT的冰体环向裂纹尺寸识别方法由周利，等创作，采用知识共享署名4.0国际许可协议进行许可。

摘要

摘要: 目的在使用环向裂纹法计算冰载荷时，裂纹的尺寸都是采用经验公式的方法进行估算，在选择经验参数时往往存在很多不确定性，为此提出一种基于YOLACT的环向裂纹参数计算方法，以准确获取海冰裂纹尺寸。方法通过YOLACT网络识别随船拍摄图像中的冰块，然后对检测到的掩膜进行边缘检测，得到冰块裂纹形状，对裂纹进行圆弧拟合获取半径和张角。结果计算结果表明，半径的识别准确率达96.12%，张角的识别准确率达96.58%。结论该方法可在基于环向裂纹法计算冰载荷时，为环向裂纹法提供裂纹尺寸的精确输入，有助于极地船舶和寒冷地区海洋结构物的初始设计。
- 环向裂纹尺寸 /
- 实例分割 /
- 海冰图像处理 /
- 圆弧拟合
Abstract: Objectives When using the circumference crack method to calculate the ice load, the size of the circumference crack is usually determined by an empirical formula, but this can introduce many uncertainties to the selection of input parameters. Methods To accurately determine the size of circumference cracks, we propose a method of recognizing ice circumference crack size based on YOLACT, which can identify ice in images taken with ships. We then detect the edge of the mask and obtain the shape of the ice cracks, and the ice breaking radius and angle can be estimated by fitting the cracks. Results The results show that the accuracy of the ice breaking radius and open angle estimated by the proposed method can reach up to 96.12% and 96.58% respectively. Conclusions This method can accurately determine the size of circumference cracks and assist in the initial design of marine structures in cold regions.
- circumference crack size /
- instance segmentation /
- sea ice image processing /
- arc fitting

HTML全文

图 1 YOLACT框架图

Figure 1. Network of YOLACT

下载: 全尺寸图片幻灯片

图 2 YOLACT检测结果

Figure 2. Test of YOLACT

下载: 全尺寸图片幻灯片

图 3 ROI二值化

Figure 3. Binarization of ROI

下载: 全尺寸图片幻灯片

图 4 原图边缘提取结果

Figure 4. Result of edge extraction in original image

下载: 全尺寸图片幻灯片

图 5 海冰边缘提取结果

Figure 5. Extraction result of sea ice edge

下载: 全尺寸图片幻灯片

图 6 圆弧端点示意图

Figure 6. Schematic diagram of arc endpoint

下载: 全尺寸图片幻灯片

图 7 中间点选取示意图

Figure 7. Schematic diagram of middle point selection

下载: 全尺寸图片幻灯片

图 8 环向裂纹参数计算流程

Figure 8. Calculation process of circumferential crack parameters

下载: 全尺寸图片幻灯片

图 9 弧线拟合结果

Figure 9. Result of arc fitting

下载: 全尺寸图片幻灯片

图 10 破冰半径计算结果

Figure 10. Calculation of ice breaking radius

下载: 全尺寸图片幻灯片

图 11 本文改进算法和霍夫圆检测算法拟合结果比较

Figure 11. Comparison of fitting results between Hough circle detection and the improved algorithm

下载: 全尺寸图片幻灯片

表 1 试验冰环向裂纹参数计算

Table 1. Calculation of circumferential crack parameters of test ice

冰块编号	破冰半径/（pixel）	破冰角/(°)
1	333	75
2	135	44
3	145	88
4	121	58
5	39	84
6	131	68
7	338	65
8	73	47

下载: 导出CSV

表 2 环向裂纹参数计算准确率

Table 2. Accuracy of calculated circumferential crack parameters

冰块编号	拟合半径/（pixel）	拟合破冰角/（°）	霍夫圆检测半径/（pixel）	霍夫圆检测角度/（°）	半径准确率/%	破冰角准确率/%
1	333	75	338	73	98.52	97.26
2	135	44	147	42	91.84	95.24
3	145	88	146	89	99.32	98.88
4	121	58	131	55	92.37	94.55
5	187	87	192	86	97.40	98.85
6	179	45	187	42	98.72	93.33
7	131	68	132	69	99.24	98.55
8	338	65	341	64	99.12	98.44
9	73	47	79	44	92.41	93.18
10	425	58	426	57	99.77	98.28

下载: 导出CSV

参考文献(15)

[1]	WANG M F, SU J, LANDY J, et al. A new algorithm for sea ice melt pond fraction estimation from high‐resolution optical satellite imagery[J]. Journal of Geophysical Research: Oceans, 2020, 125(10): e2019JC015716. doi: 10.1029/2019JC015716
[2]	倪宝玉, 徐莹, 黄其, 等. 修正内聚区长度计算公式在冰I型裂纹扩展中的应用[J]. 中国舰船研究, 2022, 17(3): 58–66. doi: 10.19693/j.issn.1673-3185.02701 NI B Y, XU Y, HUANG Q, et al. Application of improved cohesive zone length formula in ice mode I crack propagation[J]. Chinese Journal of Ship Research, 2022, 17(3): 58–66 (in Chinese). doi: 10.19693/j.issn.1673-3185.02701
[3]	丁仕风, 蔡金延, 周利, 等. 冰水池结冰过程的数值模拟分析[J]. 中国舰船研究, 2021, 16(5): 137–142. doi: 10.19693/j.issn.1673-3185.02113 DING S F, CAI J Y, ZHOU L, et al. Numerical simulation analysis of icing process in ice model tank[J]. Chinese Journal of Ship Research, 2021, 16(5): 137–142 (in Chinese). doi: 10.19693/j.issn.1673-3185.02113
[4]	BOURGAULT D. Shore-based photogrammetry of river ice[J]. Canadian Journal of Civil Engineering, 2008, 35(1): 80–86. doi: 10.1139/L07-087
[5]	卢鹏, 李志军, 哈斯·克里斯蒂安. 从船侧倾斜拍摄图像中提取海冰密集度的方法[J]. 大连海事大学学报, 2009, 35(2): 15–18. doi: 10.16411/j.cnki.issn1006-7736.2009.02.029 LU P, LI Z J, HASS C. A method to obtain sea ice concentration from oblique images on shipboard[J]. Journal of Dalian Maritime University, 2009, 35(2): 15–18 (in Chinese). doi: 10.16411/j.cnki.issn1006-7736.2009.02.029
[6]	王明锋, 苏洁, 李涛, 等. 基于无人机观测的北极冰面融池及冰面粗糙度信息提取方法研究[J]. 极地研究, 2017, 29(4): 436–445. doi: 10.13679/j.jdyj.2017.4.436 WANG M F, SU J, LI T, et al. Study on the method of extracting Arctic melt pond and roughness information on sea ice surface based on UAV observation[J]. Chinese Journal of Polar Research, 2017, 29(4): 436–445 (in Chinese). doi: 10.13679/j.jdyj.2017.4.436
[7]	WU X, XU L. Ice flood velocity calculating approach based on single view metrology[C]//IOP Conference Series: Earth and Environmental Science. Beijing: IOP, 2017: 012006.
[8]	ZHANG Q, VAN DER WERFF S, METRIKIN I, et al. Image processing for the analysis of an evolving broken-ice field in model testing[C]//International Conference on Ocean, Offshore and Arctic Engineering. Rio de Janeiro: American Society of Mechanical Engineers, 2012: 597-605.
[9]	REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once: unified, real-time object detection[C]//Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas: IEEE, 2016: 779-788.
[10]	BOLYA D, ZHOU C, XIAO F Y, et al. YOLACT++ better real-time instance segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022, 44(2): 1108–1121. doi: 10.1109/TPAMI.2020.3014297
[11]	ZHOU L, CHUANG Z J, BAI X. Ice forces acting on towed ship in level ice with straight drift. Part II: Numerical simulation[J]. International Journal of Naval Architecture and Ocean Engineering, 2018, 10(2): 119–128. doi: 10.1016/j.ijnaoe.2017.06.001
[12]	WANG S Q. A dynamic model for breaking pattern of level ice by conical structures[J]. Acta Polytechnica Scandinavica:Mechanical Engineering Series, 2001(156): 2+6–94.
[13]	SU B, RISKA K, MOAN T. Numerical simulation of local ice loads in uniform and randomly varying ice conditions[J]. Cold Regions Science and Technology, 2011, 65(2): 145–159. doi: 10.1016/j.coldregions.2010.10.004
[14]	ZHANG C Q, CHEN X D, JI S Y. Semantic image segmentation for sea ice parameters recognition using deep convolutional neural networks[J]. International Journal of Applied Earth Observation and Geoinformation, 2022, 112: 102885. doi: 10.1016/j.jag.2022.102885
[15]	SUZUKI S, BE K. Topological structural analysis of digitized binary images by border following[J]. Computer Vision, Graphics, and Image Processing, 1985, 30(1): 32–46. doi: 10.1016/0734-189X(85)90016-7