基于深度卷积特征的水下目标智能识别方法

杜 雪; 廖泓舟; 张 勋

doi:10.11993/j.issn.2096-3920.2019.03.004

基于深度卷积特征的水下目标智能识别方法

doi: 10.11993/j.issn.2096-3920.2019.03.004

哈尔滨工程大学海洋装置与控制技术研究所, 黑龙江哈尔滨, 150001

基金项目: 国家青年自然科学基金项目资助(51709062); 哈尔滨市科技创新人才研究专项基金资助(2017RAQXJ150); 中央高校基本科研专项基金(HEUCFM180402)

详细信息

作者简介:
杜雪(1987-), 女, 博士, 副教授, 主要研究方向为水下无人航行器协同导航与智能控制.

中图分类号: TJ6; TP391.41; TP183
计量
- 文章访问数: 1021
- HTML全文浏览量: 1
- PDF下载量: 579
- 被引次数: 0
出版历程
- 收稿日期: 2018-09-30
- 修回日期: 2018-12-25
- 刊出日期: 2019-06-30

Underwater Target Recognition Method Based on Deep Convolution Feature

Marine Equipment and Control Technology Research Institute, Harbin Engineering University, Harbin 150001, China

摘要

摘要: 无人水下航行器(UUV)由于本身的便利性和自主性在水下探测方面相比传统探测具有很大优势, 对UUV水下目标智能识别方法的研究具有重要意义。针对水下环境的高噪声、低对比度的特点, 文中首先使用中值滤波和局部增强处理进行水下图像预处理, 基于水下图像的小样本特点, 提出借鉴牛津大学视觉几何组网络（VGGNet)的逐层递增卷积层思想, 利用深度卷积神经网络(DCNN)设计水下智能识别框架并利用大数据集Cifar-10进行一次训练, 以学习图像通用特征; 同时使用迁移学习和数据增强技术进行二次学习, 以学习水下目标特有特征, 解决水下数据集不足的情况, 防止过拟合。通过仿真试验进行识别效果验证, 仿真结果表明, 在特定测试集下提出的水下目标智能识别方法在识别效果与自动化程度方面相比传统识别算法具有明显优势。
- 无人水下航行器 /
- 智能识别 /
- 深度卷积神经网络 /
- 迁移学习
Abstract: With the help of artificial intelligence technology, the intelligent recognition technology for unmanned undersea vehicle(UUV) to detect underwater target is presented. According to the characteristics of high noise and low contrast in underwater environment, median filtering and local enhancement are adopted to preprocess underwater images. Based on the characteristics of small sample of underwater images, the idea of incremental convolution layer by layer is proposed drawing from the visual geometry group net(VGGNet), and an underwater intelligent recognition framework is designed by using deep convolution neural network (DCNN) for training on the large data set Cifar-10 and learning general features of the image. In addition, migration learning and data enhancement technology are used for transferring knowledge to learn the unique features of underwater targets, so as to solve the problem of insufficient underwater data sets and to prevent over-fitting. Simulation results show that the proposed underwater target intelligent recognition technology has obvious advantages over the traditional algorithm in recognition effect and automation degree on specific test set.
- unmanned undersea vehicle(UUV) /
- intelligent recognition /
- deep convolution neural network(DCNN) /
- migration learning

HTML全文

参考文献(18)

[1]	Zhang W K, Wang G X, Xu G H, et al. Development of Control System in Abdominal Operating ROV[J]. Chinese Journal of Ship Research, 2017, 12(2): 124-132.
[2]	Apgar J F, Edelson J S, Furtado J, et al. ORCA-IX: An Autonomous Underwater Vehicle[D]. Massachusetts In-stitute of Technology. Massachusetts, US: Massachusetts Institute of Technology, 2006.
[3]	Zhang L, Jiang D P, Zhao J X. The Basic Control System of an Ocean Exploration AUV[J]. Applied Mechanics and Materials, 2013(411/412/413/414): 1757-1761.
[4]	Ferri G, Munafo A, Lepage K D. An Autonomous Underwater Vehicle Data-Driven Control Strategy for Target Tracking[J]. IEEE Journal of Oceanic Engineering, 2018, 99: 1-21.
[5]	Jaffe M. Underwater Optical Imaging: Status and Prospects[J]. Oceanography, 2001, 14(3): 66-76.
[6]	Lin Y, Lv F, Zhu S, et al. Large-scale Image Classification: Fast Feature Extraction and SVM Training[C]//IEEE Conference on Computer Vision and Pattern Recognition, Providence. Colorado: IEEE, 2011: 1689-1696.
[7]	Wolek A, Dzikowicz B R, Mcmahon J, et al. At-Sea Evaluation of an Underwater Vehicle Behavior for Passive Target Tracking[J]. IEEE Journal of Oceanic Engineering, 2018, 99: 1-10.
[8]	Kim J, Cho H, Pyo J, et al. The Convolution Neural Network Based Agent Vehicle Detection Using Forward-looking Sonar Image[C]//2016’ IEEE/OES Oceans. Shanghai: IEEE, 2016: 1-5.
[9]	Matias V T. End-to-end Object Detection and Recognition in Forward-looking Sonar images with Convolutional Neural Networks[C]//2016’ IEEE/OES Autonomous Underwater Vehicles. Tokyo: IEEE, 2016: 144- 150.
[10]	Qin H, Li X, Liang J, et al. Deep Fish: Accurate Underwater Live Fish Recognition with a Deep Architecture[J]. Neurocomputing, 2016, 187: 49-58.
[11]	Li X, Shang M, Hao J, et al. Accelerating Fish Detection and Recognition by Sharing CNNs with Objectness Learning [C]//2016’ IEEE/OES Oceans. Shanghai: IEEE, 2016: 1-5.
[12]	Zhu P, Isaacs J, Fu B, et al. Deep Learning Feature Extraction for Target Recognition and Classification in Underwater Sonar Images[C]//2017’ IEEE 56th Annual Conference on Decision and Control (CDC). Melbourne: IEEE, 2017: 2724-2731.
[13]	王强, 曾向阳. 深度学习方法及其在水下目标识别中的应用[C]//中国声学学会水声学分会2015年学术会议论文集. 武汉: 中国声学学会, 2015.
[14]	Zuiderveld K. Contrast Limited Adaptive Histogram Equalization[M]. Graphic Gems IV. San Diego: Academic Press Professional, 1994: 474-485.
[15]	Pan S J, Yang Q. A Survey on Transfer Learning[J]. IEEE Transactions on Knowledge & Data Engineering, 2010, 22(10): 1345-1359.
[16]	Cheng P M, Malhi H S. Transfer Learning with Convolutional Neural Networks for Classification of Abdominal Ultrasound Images[J]. Journal of Digital Imaging, 2017, 30(2): 234.
[17]	龙明盛. 迁移学习问题与方法研究[D]. 北京: 清华大学, 2014.
[18]	Li X X, Deng Z H, Choi K S. Generalized Hidden-Mapping Transductive Transfer Learning for Recognition of Epileptic Electroencephalogram Signals[J]. IEEE Transactions on Cybernetics, 2018, 99: 1-15.