Page 146 - 《应用声学》2021年第1期

第 40 卷 第 1 期 Vol. 40, No. 1
2021 年 1 月 Journal of Applied Acoustics January, 2021

⋄ 研究报告 ⋄

改进次最佳检测在侧扫声呐底混响抑制中的应用 ∗

马龙双 1,2 许 枫 1† 刘 佳 1 蒋立军 1

(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)

摘要：侧扫声呐进行沉底小目标探测时，底混响是主要背景干扰。底混响通常是一种非平稳、非高斯的带限噪
声，它使得白噪声条件下的滤波器性能受到限制。在混响背景下常利用自回归模型对接收信号进预行白化处
理，但对于实际侧扫声呐应用，白化后直接匹配滤波的处理效果不甚理想。针对此问题，在自回归模型预白化
的基础上，提出采用一种次最佳检测与多分辨二分奇异值分解相结合的改进方法。该方法首先对接收信号进
行分段处理，利用改进 Burg 算法估计每段数据自回归模型的系数及阶数；然后构造白化滤波器对分段数据预
白化，并对白化后的数据进行多分辨二分奇异值分解；最后应用 ostu 方法对原始声图和处理后的声图进行目
标检测。仿真与实验结果表明，该方法明显提高了信混比，改善了侧扫声呐沉底静态小目标的成图质量，有利
于后期实现基于图像的目标自动检测。
关键词：侧扫声呐；底混响；自回归模型；多分辨二分奇异值分解
中图法分类号: O427.9 文献标识码: A 文章编号: 1000-310X(2021)01-0142-07
DOI: 10.11684/j.issn.1000-310X.2021.01.017
Application of improved sub-optimal detection in the side-scan sonar bottom

reverberation suppression

MA Longshuang 1,2 XU Feng 1 LIU Jia 1 JIANG Lijun 1

(1 Institute of Acoustics, Chinese Academy of Sciences, Beijing 100190, China)
(2 University of Chinese Academy of Sciences, Beijing 100049, China)

Abstract: Bottom reverberation is the main background interference of side-scan sonar target detection. It is
often considered as a kind of non-stationary colored noise, which degrades the performance of the matched filter
that is the optimum detector in white background noise. The received signals are used to be pre-whitened with
the auto-regressive (AR) model in reverberation background, however, in actual side-scan sonar applications,
the result is not satisfying under the background of strong reverberation. In order to solve the problem, based on
the pre-whitening of the AR model, an improved method combining sub-optimal detection with multi-resolution
binary singular value decomposition (MBSVD) is proposed. The algorithm contains three steps, constructing
a whitening filter by AR model to pre-whitening the bottom reverberation, then performing multi-resolution
binary singular value decomposition on the whitened data. Finally, the original image and processed acoustic
image are processed by the ostu method. The simulation and experiment results show that the proposed
method can improve the signal-to-reverberation ratio and the acoustic image quality of side-scan sonar.
Keywords: Side-scan sonar; Bottom reverberation; Auto-regressive model; Multi-resolution binary singular
value decomposition

2020-04-26 收稿; 2020-12-09 定稿
国家自然科学基金项目 (11404365, 61801470), 中国科学院声学研究所 “青年英才计划” 项目 (QNYC201730)
∗
作者简介: 马龙双 (1988– ), 女, 河北邢台人, 博士研究生, 研究方向: 信号与信息处理。
通信作者 E-mail: xf@mail.ioa.ac.cn
†