Title: MRFCNN: the optimisation method of convolutional neural network for underwater target recognition

Authors: Hongbin Wang; Pengming Wang; Shengchun Deng; Zhenghao Gu

Addresses: College of Computer Science and Technology, Harbin Engineering University, China ' College of Computer Science and Technology, Harbin Engineering University, China ' School of Information Management and Artificial Intelligence, Zhejiang University of Finance and Economics, China ' College of Computer Science and Technology, Harbin Engineering University, China

Abstract: In the field of underwater target recognition, with the increase of various sensors, information and underwater noise, underwater target recognition is becoming more and more complicated. Therefore, traditional methods can no longer meet the current needs, and neural network has obvious advantages in dealing with the classification problems with complicated environmental information and vague background knowledge. In this paper, we discuss the optimisation problem from two aspects: feature extraction and target classification. Then we propose a correlation optimisation method based on convolutional neural network, and carry out related underwater simulation experiments. The experimental results show that the optimisation improvement has a certain improvement compared with the previous accuracy, which fully proves the effectiveness of the proposed optimisation method.

Keywords: underwater target recognition; convolutional neural network; feature extraction; target classification.

DOI: 10.1504/IJMIC.2022.10048799

International Journal of Modelling, Identification and Control, 2022 Vol.40 No.1, pp.36 - 43

Received: 16 Jul 2021
Accepted: 01 Sep 2021

Published online: 12 Jul 2022 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article