Page 126 - 《应用声学》2021年第2期

第 40 卷 第 2 期 Vol. 40, No. 2
2021 年 3 月 Journal of Applied Acoustics March, 2021

⋄ 研究报告 ⋄

基于形状特征的水声图像小目标识别方法

巩文静 1,2,3 田 杰 1,3† 黄海宁 1,3

(1 中国科学院声学研究所 北京 100190)
(2 中国科学院大学 北京 100049)
(3 中国科学院先进水下信息技术重点实验室 北京 100190)
摘要：为了抑制背景噪声，提高目标识别准确率，该文提出一种基于形状特征的水声图像小目标识别方法。对
含有目标的水声图像进行非局部均值去噪处理后，使用 OTSU 算法自适应选取阈值对去噪图像进行二值化分
割，结合形态学处理获得分割后的目标区域；提取目标区域的矩形度、圆形度、几何不变矩等各项形状参数，将
目标的特征向量输入随机森林分类器实现对目标形状的识别。在仿真和实测数据集上分别进行了实验，结果
表明，该方法对水声图像中的目标具有较高的识别率，可以实现不同高斯噪声背景下的目标识别，相较于其他
方法在识别率上有一定提高。
关键词：水声图像；OTSU 算法；形状特征；不变矩；目标识别
中图法分类号: TB566 文献标识码: A 文章编号: 1000-310X(2021)02-0294-09
DOI: 10.11684/j.issn.1000-310X.2021.02.018

Underwater sonar image small target recognition method based on shape features

GONG Wenjing 1,2,3 TIAN Jie 1,3 HUANG Haining 1,3

(1 Institute of Acoustics, Chinese Academy of Sciences, Beijing 100190, China)
(2 University of Chinese Academy of Sciences, Beijing 100049, China)
(3 Key Laboratory of Science and Technology on Advanced Underwater Acoustic Signal Processing, Chinese Academy of
Sciences, Beijing 100190, China)

Abstract: In order to suppress background noise and improve the accuracy of target recognition, an underwater
sonar image small target recognition method based on shape features is proposed in this paper. After performing
non-local mean denoising on the underwater sonar image containing the target, the OTSU algorithm and
morphological processing are used to adaptively select the threshold to perform binarization segmentation
of the denoised image and obtain the target area; extract the rectangularity, circularity, geometric invariant
moments and other shape parameters of the target area, input the feature vector into the random forest classifier
to recognize the shape of target. Experiments have been carried out on simulation and measured data sets. The
results show that this method has high recognition rate for targets in underwater sonar images, and can achieve
target recognition under different Gaussian noise backgrounds. There is a certain improvement compared with
other methods on recognition rate.
Keywords: Underwater sonar image; OTSU algorithm; Shape feature; Invariant moment; Target recognition

2020-06-18 收稿; 2020-09-03 定稿
作者简介: 巩文静 (1996– ), 女, 山东泰安人, 硕士研究生, 研究方向: 信号与信息处理。
† 通信作者 E-mail: tianjie@mail.ioa.ac.cn