Page 245 - 应用声学2019年第4期

第 38 卷 第 4 期 Vol. 38, No. 4
2019 年 7 月 Journal of Applied Acoustics July, 2019

⋄ 李启虎院士八十华诞学术论文 ⋄

基于成像声呐DIDSON的水域内鱼群数量

估计方法 ∗

荆丹翔 1 周晗昀 1 韩 军 1† 张 进 2
(1 浙江大学海洋学院 舟山 316021)
(2 上海海洋大学海洋科学学院 上海 201306)

摘要 为准确估计整片水域中的鱼群数量，提出一种利用成像声呐进行数量估计的方法。将成像声呐固定在
调查船下，并使波束发射方向与船前进方向一致，通过走航调查方式采集水下信息，对采集的数据进行声呐图
像构建、噪声去除、目标提取，其中噪声去除采用固定数据窗口的迭代最小二乘法，目标提取采用基于三倍标
准差准则的阈值分割法。接着利用扩展卡尔曼滤波结合最近邻的多目标跟踪算法对图像中的个体目标进行一
一计数，同时统计声呐扫描过的水域面积，获得目标个数的平均面密度值，最后结合水域占地面积，估算出整
片水域中的鱼群数量。利用该方法实现对滴水湖鱼群数量的估计，通过与人工计数结果比较，发现基于声呐图
像处理的数量统计方法具有较高精度，两者的统计值相差约 10%。
关键词 成像声呐，数量估计，噪声去除，目标提取，目标跟踪
中图法分类号: S932.4 文献标识码: A 文章编号: 1000-310X(2019)04-0705-07
DOI: 10.11684/j.issn.1000-310X.2019.04.030

Fish abundance estimation based on an imaging sonar

JING Danxiang 1 ZHOU Hanyun 1 HAN Jun 1 ZHANG Jin 2

(1 Ocean College, Zhejiang University, Zhoushan 316021, China)
(2 College of Marine Sciences, Shanghai Ocean University, Shanghai 201306, China)
Abstract In order to estimate the fish abundance accurately in a body of water, a method of fish quantity
estimation using an imaging sonar is proposed. The imaging sonar is fixed under the survey ship, making the
beam lunching direction the same with the ship’s moving direction, and sonar data is collected by means of the
investigation on navigation. The image reconstruction, noise reduction and target extraction are conducted in
turn. Fixed data window recursive least squares (FDWRLS) algorithm is adopted to reduce the speckle noise in
sonar images, and threshold segmentation method based on thrice standard error principle is utilized to extract
multiple targets. Therefore, multiple fish targets can be tracked using the nearest neighbor (NN) algorithm
combined with extended Kalman filtering (EKF) and the targets are counted one by one. Meanwhile, the area
detected by the sonar is added together to obtain the average areal density of fish. Finally, the fish abundance
in this body of water is estimated with the parameter of water area. A field experiment is conducted in Dishui
Lake to validate the effectiveness of the proposed method. Manual counting results of two datasets picked up
from the sonar data are compared with the ones by the proposed method, it is shown that the fish abundance
estimation method based on sonar image processing provides a higher accuracy rate, and the deviation between
them is approximately 10%.
Key words Imaging sonar, Abundance estimation, Noise reduction, Target extraction, Target tracking

2019-02-10 收稿; 2019-03-25 定稿
浙江省自然科学基金项目 (LY17C190007)
∗
作者简介: 荆丹翔 (1990- ), 男, 江苏镇江人, 博士研究生, 研究方向: 水声信号处理。
† 通讯作者 E-mail: jhan@zju.edu.cn