自主式水下机器人（AUV）是应用于复杂海洋环境中的高智能化无人装备，其需要具备良好的环境感知能力进行自主导航，包括水下目标识别能力。随着人工智能的高速发展，卷积神经网络作为图像处理领域的深度学习架构，在图像特征提取和图像识别上有着强大的性能和卓越的优势。本文利用卷积神经网络，实现了自主式水下机器人水下目标的自主识别。同时，通过采用三段式全连接方式和增加卷积层深度的方式对卷积神经网络进行进一步改进，提高了卷积神经网络的训练速度、准确率和泛化能力。
Autonomous underwater vehicle(AUV) is working in complex marine environment as a highly intelligent unmanned equipment. It must be provided with the key ability of the environment perception for autonomous navigation, including underwater target recognition. With the rapid development of artificial intelligence, Convolutional Neural Network(CNN), as a deep learning architecture in the field of image processing, has strong performance and superior advantages in image feature extraction and image recognition. In this paper, by the application of CNN in AUV, the autonomous underwater target recognition is realized. At the same time, further improvement of CNN is provided, by using three-stage fully connection and increasing the depth of convolution neural network, and the training speed, accuracy and generalization of CNN are improved.
2021,43(4): 155-158 收稿日期：2019-11-15
DOI：10.3404/j.issn.1672-7649.2021.04.031
分类号：TP273
作者简介：李昱(1994-),女,硕士研究生,主要研究方向为计算机视觉
参考文献：
[1] 钟宏伟. 国外无人水下航行器装备与技术现状及展望[J]. 水下无人系统学报, 2017, (4)
[2] 段世梅, 康凤举, 王彦恺. 基于SINS/DVL/GPS的AUV组合导航技术[J]. 火力与指挥控制, 2009, 34(12): 50-52
[3] Murat KÜÇÜKBAYRAK, Özhan GÜNEŞ, Arica N. Underwater Acoustic Signal Recognition Methods[J]. 2009.
[4] SHIN H C, ROTH H R, GAO M, et al. Deep convolutional neural networks for computer-aided detection: CNN architectures, dataset characteristics and transfer learning[J]. IEEE Transactions on Medical Imaging, 2016, : 1-1
[5] RUSSELL S J, NORVIG P, et al. Instructor’s manual: exercise solutions for artificial intelligence a modern approach[J]. Second Edition. Artificial Intelligence A Modern Approach, 2015, 15(96): 217-218
[6] HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition[C]. IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas: Computer Science, 2016: 272–275.
[7] BOUVRIE J. Notes on convolutional neural networks, MIT CBCL Tech Report, Cambridge, MA, 2006.
[8] 黄德双. 神经网络模式识别系统理论[M]. 北京: 电子工业出版社, 1996.
