RepDNet: A re-parameterization despeckling network for autonomous underwater side-scan sonar imaging with prior-knowledge customized convolution

• Published in: Defence technology Vol. 35; no. 5; pp. 259 - 274
• Main Authors: Li, Zhuoyi, Wang, Zhisen, Chen, Deshan, Yip, Tsz Leung, Teixeira, Angelo P.
• Format: Journal Article
• Language: English
• Published: Beijing Elsevier B.V 01.05.2024; KeAi Publishing Communications Ltd; School of Transportation and Logistics Engineering,Wuhan University of Technology,China%State Key Laboratory of Maritime Technology and Safety,Wuhan University of Technology,China; Department of Logistics and Maritime Studies,The Hong Kong Polytechnic University,China%State Key Laboratory of Maritime Technology and Safety,Wuhan University of Technology,China; National Engineering Research Center for Water Transport Safety,Wuhan University of Technology,China%Centre for Marine Technology and Ocean Engineering(CENTEC),Instituto Superior Técnico,Universidade de Lisboa,Portugal; School of Transportation and Logistics Engineering,Wuhan University of Technology,China; KeAi Communications Co., Ltd
• Subjects: Acoustics; Artificial neural networks; Autonomous underwater vehicles; Datasets; Deep learning; Domain knowledge; Energy consumption; Image processing; Inference; Methods; Noise; Noise reduction; Ocean bottom; Ocean floor; Parameterization; Re-parameterization; Side scan sonar; Signal processing; Smoothing; Sonar; Sonar image despeckling; Underwater acoustics; Underwater exploration; Sonar image despeckling; Domain knowledge; Re-parameterization; Side-scan sonar
• ISSN: 2214-9147; 2096-3459; 2214-9147
• DOI: 10.1016/j.dt.2023.12.007

Side-scan sonar (SSS) is now a prevalent instrument for large-scale seafloor topography measurements, deployable on an autonomous underwater vehicle (AUV) to execute fully automated underwater acoustic scanning imaging along a predetermined trajectory. However, SSS images often suffer from speckle noise caused by mutual interference between echoes, and limited AUV computational resources further hinder noise suppression. Existing approaches for SSS image processing and speckle noise reduction rely heavily on complex network structures and fail to combine the benefits of deep learning and domain knowledge. To address the problem, RepDNet, a novel and effective despeckling convolutional neural network is proposed. RepDNet introduces two re-parameterized blocks: the Pixel Smoothing Block (PSB) and Edge Enhancement Block (EEB), preserving edge information while attenuating speckle noise. During training, PSB and EEB manifest as double-layered multi-branch structures, integrating first-order and second-order derivatives and smoothing functions. During inference, the branches are re-parameterized into a 3 × 3 convolution, enabling efficient inference without sacrificing accuracy. RepDNet comprises three computational operations: 3 × 3 convolution, element-wise summation and Rectified Linear Unit activation. Evaluations on benchmark datasets, a real SSS dataset and Data collected at Lake Mulan aestablish RepDNet as a well-balanced network, meeting the AUV computational constraints in terms of performance and latency.