In response to the current challenges in efficiently acquiring sound speed profiles and ensuring their representativeness, considering the need to fully leverage historical sound speed profiles while accounting for their spatiotemporal variability, we introduce a model for sound speed profile prediction based on a CNN-BiLSTM-Attention network, which integrates a convolutional neural network (CNN), a bidirectional long short-term memory network (BiLSTM), and an attention mechanism (AM). The synergy of these components enables the model to extract the spatiotemporal features of sound speed profiles more comprehensively. Utilizing the global ocean Argo grid dataset, the model predicted the sound speed profiles of an experimental zone in the Western Pacific Ocean. In predicting sound speed profiles of a single point, the model achieved a root mean square error (RMSE), relative error (RE), and accuracy (ACC) of 0.72 m/s, 0.029%, and 0.99971, respectively, surpassing comparative models. For regional sound speed profile prediction, the mean RMSE, RE, and ACC of different water layers were 0.919 m/s, −0.016%, and 0.9995, respectively. The experimental outcomes not only confirm the high accuracy of the model, but also highlight its superiority in sound speed profile prediction, particularly as an effective compensatory approach when profile measurements are untimely or contain representational errors.

Given the lack of systematic research on bathymetric surveys with multi-beam sonar carried by autonomous underwater vehicles (AUVs) in unfamiliar waters, this paper proposes a method for multi-beam bathymetric surveys based on the constant-depth mode of AUVs, considering equipment safety, operational efficiency, and data quality. Firstly, basic principles for multi-beam bathymetric surveys under the constant-depth mode are proposed based on multi-beam operational standards and AUV constant-depth mode characteristics. Secondly, a vertical effective height model for the vehicle is established, providing vertical constraints and a basis for determining fixed depth in constant-depth missions. Subsequently, according to these basic principles and the vertical effective height model, the operational process for multi-beam bathymetric surveys in unfamiliar waters under the AUV constant-depth mode is outlined. Finally, we validate the proposed method through sea trials in the Xisha Sea of the South China Sea. The test results show that the method proposed in this paper not only ensures the vehicle safety operation and multi-beam data quality, but also improves the operation efficiency by about 68%, demonstrating the reliability of the proposed method and its significant engineering value and guidance implications.

The scarcity and difficulty in acquiring Side-scan sonar target images limit the application of deep learning algorithms in Side-scan sonar target detection. At present, there are few amplification methods for Side-scan sonar images, and the amplification image quality is not ideal, which is not suitable for the characteristics of Side-scan sonar images. Addressing the current shortage of sample augmentation methods for Side-scan sonar, this paper proposes a method for augmenting single underwater target images using the CBL-sinGAN network. Firstly, considering the low resolution and monochromatic nature of Side-scan sonar images while balancing training efficiency and image diversity, a sinGAN network is introduced and designed as an eight-layer pyramid structure. Secondly, the Convolutional Block Attention Module (CBAM) is integrated into the network generator to enhance target learning in images while reducing information diffusion. Finally, an L1 loss function is introduced in the network discriminator to ensure training stability and improve the realism of generated images. Experimental results show that the accuracy of shipwreck target detection increased by 4.9% after training with the Side-scan sonar sample dataset augmented by the proposed network. This method effectively retains the style of the images while achieving diversity augmentation of small-sample underwater target images, providing a new approach to improving the construction of underwater target detection models.

The speed of sound in seawater plays a crucial role in determining the accuracy of multibeam bathymetric measurements. In deep-sea multibeam measurements, the challenge of inadequate longitudinal coverage of sound speed profiles arises from variations in seafloor topography, meteorological conditions, measurement equipment, and operational efficiency, resulting in diminished measurement precision. Building upon the EOF (Empirical Orthogonal Function), a method employed to analyze spatiotemporal data such as sound speeds, this paper addresses the limitations of the EOF method caused by the shallowest sampling depth of the sound speed profile samples. It proposes two methods for EOF reconstruction of measured sound speed profiles extended to full water depth by splicing measured sound speed profiles at non-full water depths with historical average sound speed profiles of the surveyed sea area. Specially, Method 2 introduces the latest metaheuristic optimization algorithm, CPO (Crested Porcupine Optimizer), which exhibited superior performance on multiple standard test functions in 2024. The study reconstructs randomly sampled measured sound speed profiles using the two proposed methods and commonly employed substitution and splicing methods, followed by a comparative analysis of the experimental outcomes. At a sampling depth of 200 m, Method 2 demonstrates performance superior to other methods, with RMSE, MAE, MAPE, and R² values of 0.9511 m/s, 0.8492 m/s, 0.0566%, and 0.9963, respectively. Method 1 yields corresponding values of 0.9594 m/s, 0.8492 m/s, 0.0568%, and 0.9962, respectively. Despite its slightly inferior performance compared with Method 2, it offers substantial advantages over the substitution and splicing methods. Varying the sampling depth of measured sound speed profiles reveals that Methods 1 and 2 exhibit inferior reconstruction performance in shallow water compared with the substitution and splicing methods. Nevertheless, when the sampling depth surpasses the depth range of initial spatial modes with abrupt variations, both methods achieve notably higher reconstruction accuracy compared with the substitution and splicing methods, reaching a stabilized state. Sound ray tracing reveals that the reconstructed sound speed profiles from both methods meet the stringent accuracy standards for bathymetric measurements, achieving an effective beam ratio of 100%. The proposed methods not only provide rapid reconstruction of sound speed profiles, thereby improving the efficiency of multibeam bathymetric surveys, but also provide references for the reasonable determination of sampling depths of sound speed profiles to ensure reconstruction accuracy, demonstrating practical application value.

Super-resolution (SR) is a technique that restores image details based on existing information, enhancing the resolution of images to prevent quality degradation. Despite significant achievements in deep-learning-based SR models, their application in underwater sonar scenarios is limited due to the lack of underwater sonar datasets and the difficulty in recovering texture details. To address these challenges, we propose a multi-scale generative adversarial network (SIGAN) for super-resolution reconstruction of underwater sonar images. The generator is built on a residual dense network (RDN), which extracts rich local features through densely connected convolutional layers. Additionally, a Convolutional Block Attention Module (CBAM) is incorporated to capture detailed texture information by focusing on different scales and channels. The discriminator employs a multi-scale discriminative structure, enhancing the detail perception of both generated and high-resolution (HR) images. Considering the increased noise in super-resolved sonar images, our loss function emphasizes the PSNR metric and incorporates the L2 loss function to improve the quality of the output images. Meanwhile, we constructed a dataset for side-scan sonar experiments (DNASI-I). We compared our method with the current state-of-the-art super-resolution image reconstruction methods on the public dataset KLSG-II and our self-built dataset DNASI-I. The experimental results show that at a scale factor of 4, the average PSNR value of our method was 3.5 higher than that of other methods, and the accuracy of target detection using the super-resolution reconstructed images can be improved to 91.4%. Through subjective qualitative comparison and objective quantitative analysis, we demonstrated the effectiveness and superiority of the proposed SIGAN in the super-resolution reconstruction of side-scan sonar images.

In this paper, a method for augmenting samples of side-scan sonar seafloor sediment images based on CBAM-BCEL1-INGAN is proposed, aiming to address the difficulties in acquiring and labeling datasets, as well as the insufficient diversity and quantity of data samples. Firstly, a Convolutional Block Attention Module (CBAM) is integrated into the residual blocks of the INGAN generator to enhance the learning of specific attributes and improve the quality of the generated images. Secondly, a BCEL1 loss function (combining binary cross-entropy and L1 loss functions) is introduced into the discriminator, enabling it to focus on both global image consistency and finer distinctions for better generation results. Finally, augmented samples are input into an AlexNet classifier to verify their authenticity. Experimental results demonstrate the excellent performance of the method in generating images of coarse sand, gravel, and bedrock, as evidenced by significant improvements in the Frechet Inception Distance (FID) and Inception Score (IS). The introduction of the CBAM and BCEL1 loss function notably enhances the quality and details of the generated images. Moreover, classification experiments using the AlexNet classifier show an increase in the recognition rate from 90.5% using only INGAN-generated images of bedrock to 97.3% using images augmented using our method, marking a 6.8% improvement. Additionally, the classification accuracy of bedrock-type matrices is improved by 5.2% when images enhanced using the method presented in this paper are added to the training set, which is 2.7% higher than that of the simple method amplification. This validates the effectiveness of our method in the task of generating seafloor sediment images, partially alleviating the scarcity of side-scan sonar seafloor sediment image data.