端到端原则
混响
计算机科学
滤波器(信号处理)
声学
物理
计算机网络
计算机视觉
作者
Zhen Wang,Hao Zhang,Xiao Chen,Yuan An
出处
期刊:PLOS ONE
[Public Library of Science]
日期:2023-10-24
卷期号:18 (10): e0293365-e0293365
标识
DOI:10.1371/journal.pone.0293365
摘要
Reverberation is the primary background interference of active sonar systems in shallow water environments, affecting target position detection accuracy. Reverberation suppression is a signal processing technique used to improve the clarity and accuracy of echo by eliminating the echoes, reverberations, and noise that occur during underwater propagation.This paper proposes an end-to-end network structure called the Reverberation Suppression Network (RS-U-Net) to suppress the reverberation of underwater echo signals. The proposed method effectively improves the signal-to-reverberation ratio (SRR) of the echo signal, outperforming existing methods in the literature. The RS-U-Net architecture uses sonar echo signal data as input, and a one-dimensional convolutional network (1D-CNN) is used in the network to train and extract signal features to learn the main features. The algorithm's effectiveness is verified by the pool experiment echo data, which shows that the filter can improve the detection of echo signals by about 10 dB. The weights of reverberation suppression tasks are initialized with an auto-encoder, which effectively uses the training time and improves performance. By comparing with the experimental pool data, it is found that the proposed method can improve the reverberation suppression by about 2 dB compared with other excellent methods.
科研通智能强力驱动
Strongly Powered by AbleSci AI