基于分形纹理特征的侧扫声呐图像沉船识别方法研究

董凌宇; 单瑞; 刘慧敏; 于得水; 杜凯

doi:10.16562/j.cnki.0256-1492.2020070301

基于分形纹理特征的侧扫声呐图像沉船识别方法研究

1.
中国地质调查局青岛海洋地质研究所，青岛 266071

2.
青岛海洋科学与技术国家实验室海洋矿产资源评价与探测技术功能实验室，青岛 266071

基金项目: 国家自然科学基金“高频GNSS单点测速数据提取海浪参数方法研究”（41406115）；中国地质调查局项目“深海调查-测量”（DD20191003）；青岛市南区科技发展资金项目“轻便型GNSS浪潮测量浮标关键技术研究”（2016-3-015-ZH）

详细信息

作者简介: 董凌宇（1994—），男，硕士，研究实习员，主要从事地球物理数据处理与方法研究，E-mail：17663985486@163.com

通讯作者: 单瑞（1985—），男，硕士，助理研究员，主要从事海洋地球物理及海洋测绘研究工作，E-mail：shanrui416@163.com

中图分类号: P714.8

收稿日期: 2020-07-03

修回日期: 2020-08-19

刊出日期: 2021-08-28

Shipwreck identification with side scan sonar image based on fractal texture

1.
Qingdao Institute of Marine Geology, CGS, Qingdao 266071, China

2.
Laboratory for Marine Mineral Resources, Qingdao National Laboratory for Marine Science and Technology, Qingdao 266071, China

More Information

Corresponding author: SHAN Rui, shanrui416@163.com

Received Date: 03 July 2020

Revised Date: 19 August 2020

Publish Date: 28 August 2021

摘要

摘要:
为提高侧扫声呐图像中沉船等目标信息的识别精度和识别效率，根据盒维数、毯维数与多重分形谱的侧扫声呐图像纹理特征提取算法，构建了基于分形纹理特征的Adaboost级联分类器沉船目标识别流程。结合实测侧扫声呐图像数据进行水下沉船识别实验，并与灰度共生矩阵和Tamura纹理特征的识别结果进行对比。研究表明，基于分形纹理特征的识别方法综合考虑了图像全局与局部纹理特征，且不依赖人工选取阈值参数与特征向量，可有效提高目标识别精度和识别效率。
- 沉船识别 /
- 纹理特征 /
- 盒维数 /
- 毯维数 /
- 多重分形谱
Abstract:
In order to improve the accuracy and efficiency for recognition of underwater targets, fractal texture features including box dimension, blanket dimension and multifractal spectrum are calculated by texture feature extraction algorithm with side scan sonar images, and the shipwreck identification procedure based on Adaboost cascade classifier is constructed. The shipwreck recognition experiments have been carried out, and the results are compared. Research shows that the recognition method based on fractal texture features comprehensively considers the global and local texture features of the image, and does not rely on manual selection of threshold parameters and feature vectors, which can improve the accuracy and efficiency of target recognition.
- shipwreck identification /
- texture feature /
- box dimension /
- blanket dimension /
- multifractal spectrum

HTML全文

图 1 Adaboost级联分类器

Figure 1.

下载: 全尺寸图片幻灯片

图 2 基于分形纹理特征的Adaboost目标识别流程

Figure 2.

下载: 全尺寸图片幻灯片

图 3 目标识别中的正样本与负样本

Figure 3.

下载: 全尺寸图片幻灯片

图 4 盒维数计算

Figure 4.

下载: 全尺寸图片幻灯片

图 5 不同毯子厚度的分类结果比较

Figure 5.

下载: 全尺寸图片幻灯片

图 6 不同样本的多重分形谱结果

Figure 6.

下载: 全尺寸图片幻灯片

表 1 沉船及非沉船目标的多重分形谱参数

Table 1. Parameters of multifractal spectrum of shipwrecked and non-wrecked targets

目标	α_min	α_max	f_min	f_max	Δα	Δf
沉船1	1.81	2.96	0.07	2.00	1.15	1.93
沉船2	1.92	2.21	1.68	2.00	0.29	0.32
沉船3	1.90	2.35	1.13	2.00	0.45	0.87
非沉船1	1.99	2.04	1.75	2.00	0.05	0.25
非沉船2	1.98	2.05	1.75	2.00	0.07	0.25
非沉船3	1.96	2.03	1.80	2.00	0.07	0.20
注：①α_min和α_max分别代表了图像测度集的最小概率和最大概率，其差值Δα表明图像在概率测度分布中的差异程度，Δα越大则图像各测度区域和分形层次的区别越大，多重分形性质越明显；Δα越小则图像各测度区域和分形层次的区别越小，多重分形性质越微弱。 ②f_min和f_ma_x分布代表了图像测度集的最大值和最小值，其差值Δf表明图像在图像测度子集纹理复杂程度上的差异，Δf差值越大则表明图像不同测度子集纹理区别越明显。

下载: 导出CSV

表 2 分形纹理特征识别结果

Table 2. Recognition of fractal texture feature

识别方法	精确度/%	召回率/%	F₁/%
盒维数	50	78.95	61.2
毯维数	88.2	78.9	83.3
多重分形谱	95	100	97.4

下载: 导出CSV

表 3 多重分形谱、GLCM、Tamura三种纹理特征识别结果

Table 3. Recognition results of multifractal spectrum, GLCM and Tamura

识别方法	精确率/%	召回率/%	F₁/%
多重分形谱	95	100	97.4
GLCM（d=10）	100	94.7	97.2
Tamura六特征值	94.1	84.2	88.9

下载: 导出CSV

参考文献(27)

[1]	赵建虎, 王爱学. 精密海洋测量与数据处理技术及其应用进展[J]. 海洋测绘, 2015, 35(6):1-7 doi: 10.3969/j.issn.1671-3044.2015.06.001 ZHAO Jianhu, WANG Aixue. Precise marine surveying and data processing technology and their progress of application [J]. Hydrographic Surveying and Charting, 2015, 35(6): 1-7. doi: 10.3969/j.issn.1671-3044.2015.06.001
[2]	赵建虎, 王爱学, 王晓, 等. 侧扫声纳条带图像分段拼接方法研究[J]. 武汉大学学报: 信息科学版, 2013, 38(9):1034-1038 ZHAO Jianhu, WANG Aixue, WANG Xiao, et al. A segmented mosaic method for side scan sonar strip images using corresponding features [J]. Geomatics and Information Science of Wuhan University, 2013, 38(9): 1034-1038.
[3]	Rutledge J, Yuan W T, Wu J, et al. Intelligent shipwreck search using autonomous underwater vehicles[C]//2018 IEEE International Conference on Robotics and Automation (ICRA). Brisbane, QLD, Australia: IEEE, 2018: 6175-6182.
[4]	许文海, 续元君, 董丽丽, 等. 基于水平集和支持向量机的图像声呐目标识别[J]. 仪器仪表学报, 2012, 33(1):49-55 doi: 10.3969/j.issn.0254-3087.2012.01.008 XU Wenhai, XU Yuanjun, DONG Lili, et al. Level-set and SVM based target recognition of image sonar [J]. Chinese Journal of Scientific Instrument, 2012, 33(1): 49-55. doi: 10.3969/j.issn.0254-3087.2012.01.008
[5]	卞红雨, 陈奕名, 张志刚, 等. 像素重要性测量特征下的侧扫声呐目标检测[J]. 声学学报, 2019, 44(3):353-359 BIAN Hongyu, CHEN Yiming, ZHANG Zhigang, et al. Target detection algorithm in side-scan sonar image based on pixel importance measurement [J]. Acta Acustica, 2019, 44(3): 353-359.
[6]	郭军, 马金凤, 王爱学. 基于SVM算法和GLCM的侧扫声纳影像分类研究[J]. 测绘与空间地理信息, 2015, 38(3):60-63 doi: 10.3969/j.issn.1672-5867.2015.03.020 GUO Jun, MA Jinfeng, WANG Aixue. Study of side scan sonar image classification based on SVM and gray level co-occurrence matrix [J]. Geomatics & Spatial Information Technology, 2015, 38(3): 60-63. doi: 10.3969/j.issn.1672-5867.2015.03.020
[7]	高程程, 惠晓威. 基于灰度共生矩阵的纹理特征提取[J]. 计算机系统应用, 2010, 19(6):195-198 doi: 10.3969/j.issn.1003-3254.2010.06.047 GAO Chengcheng, HUI Xiaowei. GLCM-based texture feature extraction [J]. Computer Systems & Applications, 2010, 19(6): 195-198. doi: 10.3969/j.issn.1003-3254.2010.06.047
[8]	景军锋, 张缓缓, 李鹏飞, 等. LBP和Tamura纹理特征方法融合的织物疵点分类算法[J]. 计算机工程与应用, 2012, 48(23):155-160 doi: 10.3778/j.issn.1002-8331.2012.23.035 JING Junfeng, ZHANG Huanhuan, LI Pengfei, et al. Fabric defect classification based on local binary patterns and Tamura texture feature method [J]. Computer Engineering and Applications, 2012, 48(23): 155-160. doi: 10.3778/j.issn.1002-8331.2012.23.035
[9]	王瑞. 多重分形及其在图像识别中的应用研究[D]. 西北大学硕士学位论文, 2010. WANG Rui. Multifractal and its application in image recognition[D]. Master Dissertation of Northwest University, 2010.
[10]	徐文海. 基于分形理论的遥感影像纹理分析与分类研究[D]. 中南大学硕士学位论文, 2010. XU Wenhai. Texture analysis and classification of remote sensing image based on fractal theory[D]. Master Dissertation of Central South University, 2010.
[11]	Femmam S. Texture classification approach based on 2D multifractal analysis[Z]. SPIE Newsroom, 2015.
[12]	Lopes R, Betrouni N. Fractal and multifractal analysis: a review [J]. Medical Image Analysis, 2009, 13(4): 634-649. doi: 10.1016/j.media.2009.05.003
[13]	Don S, Chung D, Revathy K, et al. A neural network approach to mammogram image classification using fractal features[C]//2009 IEEE International Conference on Intelligent Computing and Intelligent Systems. Shanghai, China: IEEE, 2009: 444-447.
[14]	Cao W L, Shi Z K, Feng J H. Traffic image classification method based on fractal dimension[C]//2006 5th IEEE International Conference on Cognitive Informatics. Beijing, China: IEEE, 2006: 903-907.
[15]	李攀峰, 赵铁虎, 张晓波, 等. 山东半岛遥感解译断裂分形研究[J]. 海洋地质与第四纪地质, 2015, 35(4):105-112 LI Panfeng, ZHAO Tiehu, ZHANG Xiaobo, et al. Fractal research of remote sensing linear faults in Shandong peninsula [J]. Marine Geology & Quaternary Geology, 2015, 35(4): 105-112.
[16]	Grassberger P. Generalized dimensions of strange attractors [J]. Physics Letters A, 1983, 97(6): 227-230. doi: 10.1016/0375-9601(83)90753-3
[17]	Falconer K J. Fractal Geometry - Mathematical Foundations and Applications[M]. Chichester: Wiley, 1990.
[18]	Gagnepain J J, Roques-Carmes C. Fractal approach to two-dimensional and three-dimensional surface roughness [J]. Wear, 1986, 109(1-4): 119-126. doi: 10.1016/0043-1648(86)90257-7
[19]	Kisan S, Mishra S, Bhattacharjee G, et al. Analytical Study on Fractal Dimension-A Review[C]//2018 International Conference on Recent Innovations in Electrical, Electronics & Communication Engineering (ICRIEECE). IEEE, 2018: 380-384.
[20]	周江, 印萍, 程荡敌, 等. 基于GIS和分形理论研究的海岸线图像和分维以及长度[J]. 海洋地质与第四纪地质, 2008, 28(4):65-71 ZHOU Jiang, YIN Ping, CHENG Dangdi, et al. Research on the fractal simulation image and the fractal dimension and length of coastline based on GIS and fractal theory [J]. Marine Geology & Quaternary Geology, 2008, 28(4): 65-71.
[21]	李会方. 多重分形理论及其在图象处理中应用的研究[D]. 西北工业大学博士学位论文, 2004. LI Huifang. The study on multifractal theory and application in image processing[D]. Doctor Dissertation of Northwestern Polytechnical University, 2004.
[22]	Turiel A, Del Pozo A. Reconstructing images from their most singular fractal manifold [J]. IEEE Transactions on Image Processing, 2002, 11(4): 345-350. doi: 10.1109/TIP.2002.999668
[23]	Turiel A, Parga N. The multifractal structure of contrast changes in natural images: from sharp edges to textures [J]. Neural Computation, 2000, 12(4): 763-793. doi: 10.1162/089976600300015583
[24]	Potlapalli H, Luo R C. Fractal-based classification of natural textures [J]. IEEE Transactions on Industrial Electronics, 1998, 45(1): 142-150. doi: 10.1109/41.661315
[25]	Mahmood Z, Ali T, Khattak S. Automatic player detection and recognition in images using AdaBoost[C]//Proceedings of 2012 9th International Bhurban Conference on Applied Sciences and Technology (IBCAST). Islamabad, Pakistan: IEEE, 2012: 64-69.
[26]	李航. 统计学习方法[M]. 北京: 清华大学出版社, 2012: 18-20. LI Hang. Statistical Learning Methods[M]. Beijing: Tsinghua University Press, 2012: 18-20.
[27]	Scoville D. Steamer Homer Warren[Z/OL]. https://www.shipwreckworld.com/articles/side-scan-sonar-images/.