Search Results (155)

The ocean encompasses the majority of the Earth’s surface and harbors substantial energy resources. Nevertheless, the intricate and asymmetrically distributed underwater environment renders existing target detection performance inadequate. This paper presents an enhanced YOLOv8s approach for underwater robot object detection to address issues of subpar image quality and low recognition accuracy. The precise measures are enumerated as follows: initially, to address the issue of model parameters, we optimized the ninth convolutional layer by substituting certain conventional convolutions with adaptive deformable convolution DCN v4. This modification aims to more effectively capture the deformation and intricate features of underwater targets, while simultaneously decreasing the parameter count and enhancing the model’s ability to manage the deformation challenges presented by underwater images. Furthermore, the Triplet Attention module is implemented to augment the model’s capacity for detecting multi-scale targets. The integration of low-level superficial features with high-level semantic features enhances the feature expression capability. The original CIoU loss function was ultimately substituted with Shape IoU, enhancing the model’s performance. In the underwater robot grasping experiment, the system shows particular robustness in handling radial symmetry in marine organisms and reflection symmetry in artificial structures. The enhanced algorithm attained a mean Average Precision (mAP) of 87.6%, surpassing the original YOLOv8s model by 3.4%, resulting in a marked enhancement of the object detection model’s performance and fulfilling the real-time detection criteria for underwater robots. Full article

Underwater acoustic target recognition (UATR) is challenged by complex marine noise, scarce labeled data, and inadequate multi-scale feature extraction in conventional methods. This study proposes DART-MT, a semi-supervised framework that integrates a Dual Attention Parallel Residual Network Transformer with a mean teacher paradigm, enhanced by domain-specific prior knowledge. The architecture employs a Convolutional Block Attention Module (CBAM) for localized feature refinement, a lightweight New Transformer Encoder for global context modeling, and a novel TriFusion Block to synergize spectral–temporal–spatial features through parallel multi-branch fusion, addressing the limitations of single-modality extraction. Leveraging the mean teacher framework, DART-MT optimizes consistency regularization to exploit unlabeled data, effectively mitigating class imbalance and annotation scarcity. Evaluations on the DeepShip and ShipsEar datasets demonstrate state-of-the-art accuracy: with 10% labeled data, DART-MT achieves 96.20% (DeepShip) and 94.86% (ShipsEar), surpassing baseline models by 7.2–9.8% in low-data regimes, while reaching 98.80% (DeepShip) and 98.85% (ShipsEar) with 90% labeled data. Under varying noise conditions (−20 dB to 20 dB), the model maintained a robust performance (F1-score: 92.4–97.1%) with 40% lower variance than its competitors, and ablation studies validated each module’s contribution (TriFusion Block alone improved accuracy by 6.9%). This research advances UATR by (1) resolving multi-scale feature fusion bottlenecks, (2) demonstrating the efficacy of semi-supervised learning in marine acoustics, and (3) providing an open-source implementation for reproducibility. In future work, we will extend cross-domain adaptation to diverse oceanic environments. Full article

Underwater sonar target recognition is crucial in fields such as national defense, navigation, and environmental monitoring. However, it faces issues such as the complex characteristics of ship-radiated noise, imbalanced data distribution, non-stationarity, and bottlenecks of existing technologies. This paper proposes the MultiFuseNet-AID network, aiming to address these challenges. The network includes the TriFusion block module, the novel lightweight attention residual network (NLARN), the long- and short-term attention (LSTA) module, and the Mamba module. Through the TriFusion block module, the original, differential, and cumulative signals are processed in parallel, and features such as MFCC, CQT, and Fbank are fused to achieve deep multi-domain feature fusion, thereby enhancing the signal representation ability. The NLARN was optimized based on the ResNet architecture, with the SE attention mechanism embedded. Combined with the long- and short-term attention (LSTA) and the Mamba module, it could capture long-sequence dependencies with an O(N) complexity, completing the optimization of lightweight long sequence modeling. At the same time, with the help of feature fusion, and layer normalization and residual connections of the Mamba module, the adaptability of the model in complex scenarios with imbalanced data and strong noise was enhanced. On the DeepShip and ShipsEar datasets, the recognition rates of this model reached 98.39% and 99.77%, respectively. The number of parameters and the number of floating point operations were significantly lower than those of classical models, and it showed good stability and generalization ability under different sample label ratios. The research shows that the MultiFuseNet-AID network effectively broke through the bottlenecks of existing technologies. However, there is still room for improvement in terms of adaptability to extreme underwater environments, training efficiency, and adaptability to ultra-small devices. It provides a new direction for the development of underwater sonar target recognition technology. Full article

In order to solve problems such as recognition of blind areas which exist in traditional technology in underwater near-field target sensing, this paper constructs an underwater robot target sensing model based on the fish lateral line sensing mechanism and adopts CFD simulation technology to analyze the perturbation characteristic law of the pressure signal in the flow field around the underwater robot. By extracting the pressure signal following the bionic lateral line on the surface of the underwater robot as the target recognition information, the SVM multi-target recognition model is trained and built to realize the perception and recognition of the structural features and attitude features of the underwater robot. The results show that the structural features and attitude features of the underwater robot can be recognized by using the time-domain waveform structural features and spatially symmetric distribution features of the pressure coefficients, and the recognition accuracy can reach over 90%, which reveals the principle of target feature resolution based on the sideline perception signals of the fish nerve center. Full article

To tackle the challenges of relative attitude adaptability and limited sample availability in underwater moving target recognition for active sonar, this study focuses on key aspects such as feature extraction, network model design, and information fusion. A pseudo-three-dimensional spatial feature extraction method is proposed by integrating generalized MUSIC with range–dimension information. The pseudo-WVD time–frequency feature is enhanced through the incorporation of prior knowledge. Additionally, the Doppler frequency shift distribution feature for underwater moving targets is derived and extracted. A multidimensional feature information fusion network model based on meta-learning is developed. Meta-knowledge is extracted separately from spatial, time–frequency, and Doppler feature spectra, to improve the generalization capability of single-feature task networks during small-sample training. Multidimensional feature information fusion is achieved via a feature fusion classifier. Finally, a sample library is constructed using simulation-enhanced data and experimental data for network training and testing. The results demonstrate that, in the few-sample scenario, the proposed method leverages the complementary nature of multidimensional features, effectively addressing the challenge of limited adaptability to relative horizontal orientation angles in target recognition, and achieving a recognition accuracy of up to 97.1%. Full article

Achieving reliable target detection in the field of sonar imagery represents a significant challenge due to the complex underwater interference patterns characterized by speckle noise, tunnel effects, and low-signal-to-noise ratio (SNR) environments. Currently, constant false alarm rate (CFAR) detection denotes a fundamental target detection method in sonar target recognition. However, conventional CFAR methods face some limitations, including a slow computational speed, a high false alarm rate (FAR), and a notable missed detection rate (MDR). To address these limitations, this study proposes an innovative segmentation–detection framework. The proposed framework employs a global segmentation algorithm to identify regions of interest containing potential targets, which is followed by localized two-dimensional CFAR detection. This hierarchical framework can significantly improve computational efficiency while reducing the FAR, thus enabling the practical implementation of advanced, computationally intensive CFAR detection methods in real-time target detection in sonar imagery. In addition, an innovative segmented-ordered-weighting CFAR (SOW-CFAR) detection method that integrates multiple weighting windows to implement ordered weighting of reference cells is developed. This method can effectively reduce both the FAR and MDR through optimized reference cell processing. The experimental results demonstrate that the proposed method can achieve superior detection performance in sonar imagery applications compared to the existing methods. The proposed SOW-CFAR detection method can achieve fast and accurate target detection in the sonar imagery field. Full article

In order to improve the efficiency of the deep network model in processing the radiated noise signals of underwater acoustic targets, this paper introduces a Singular Spectrum Analysis and Channel Attention Convolutional Neural Network (SSA-CACNN) model. The front end of the model is designed as an SSA filter, and its input is the time-domain signal that has undergone simple preprocessing. The SSA method is utilized to separate the noise efficiently and reliably from useful signals. The first three orders of useful signals are then fed into the CACNN model, which has a convolutional layer set up at the beginning of the model to further remove noise from the signal. Then, the attention of the model to the feature signal channels is enhanced through the combination of multiple groups of convolutional operations and the channel attention mechanism, which facilitates the model’s ability to discern the essential characteristics of the underwater acoustic signals and improve the target recognition rate. Experimental Results: The signal reconstructed by the first three-order waveforms at the front end of the SSA-CACNN model proposed in this paper can retain most of the features of the target. In the experimental verification using the ShipsEar dataset, the model achieved a recognition accuracy of 98.64%. The model’s parameter count of 0.26 M was notably lower than that of other comparable deep models, indicating a more efficient use of resources. Additionally, the SSA-CACNN model had a certain degree of robustness to noise, with a correct recognition rate of 84.61% maintained when the signal-to-noise ratio (SNR) was −10 dB. Finally, the pre-trained SSA-CACNN model on the ShipsEar dataset was transferred to the DeepShip dataset with a recognition accuracy of 94.98%. Full article

Underwater optical imaging plays a crucial role in maritime safety, enabling reliable navigation, efficient search and rescue operations, precise target recognition, and robust military reconnaissance. However, conventional underwater imaging methods often suffer from severe backscattering noise, limited detection range, and reduced image clarity—challenges that are exacerbated in turbid waters. To address these issues, Underwater Laser Range-Gated Imaging has emerged as a promising solution. By selectively capturing photons within a controlled temporal gate, this technique effectively suppresses backscattering noise-enhancing image clarity, contrast, and detection range. Nevertheless, residual noise within the imaging slice can still degrade image quality, particularly in challenging underwater conditions. In this study, we propose an enhanced U-Net neural network designed to mitigate noise interference in underwater laser range-gated images, improving target recognition performance. Built upon the U-Net architecture with added residual connections, our network combines a VGG16-based perceptual loss with Mean Squared Error (MSE) as the loss function, effectively capturing high-level semantic features while preserving critical target details during reconstruction. Trained on a semi-synthetic grayscale dataset containing synthetically degraded images paired with their reference counterparts, the proposed approach demonstrates improved performance compared to several existing underwater image restoration methods in our experimental evaluations. Through comprehensive qualitative and quantitative evaluations, underwater target detection experiments, and real-world oceanic validations, our method demonstrates significant potential for advancing maritime safety and related applications. Full article

With the rapid advancement of ocean monitoring technology, the types and quantities of underwater sensors have increased significantly. Traditional single-sensor approaches exhibit limitations in underwater target classification, resulting in low classification accuracy and poor robustness. This paper integrates deep learning and information fusion theory to propose a multi-level fusion perception method for underwater targets based on multi-physical-field sensing. We extract both conventional typical features and deep features derived from an autoencoder and perform feature-level fusion. Neural network-based classification models are constructed for each physical field subsystem. To address the class imbalance and difficulty imbalance issues in the collected physical field target samples, we design a C-Focal Loss function specifically for the three underwater target categories. Furthermore, based on the confusion matrix results from the subsystem’s validation set, we propose a neural network-based Dempster–Shafer evidence fusion method (NNDS). Experimental validation using real-world data demonstrates a 97.15% fusion classification accuracy, significantly outperforming both direct multi-physical-field network fusion and direct subsystem decision fusion. The proposed method also exhibits superior reliability and robustness. Full article

Addressing the limitations of restricted coding capacity and material dependency in acoustic identity tags for autonomous underwater vehicles (AUVs), this study introduces a novel passive acoustic identification tag (AID) design based on multilayered elastic cylindrical shells. By developing a Normal Mode Series (NMS) analytical model and validating it through finite element method (FEM) simulations, the work elucidates how material layering strategies regulate far-field target strength (TS) and establishes a time-domain multi-peak echo-based encoding framework. Results demonstrate that optimizing material impedance contrasts achieves 99% detection success at a 3 dB signal-to-noise ratio. Jaccard similarity analysis of 3570 material combinations reveals a system-wide average recognition error rate of 0.41%, confirming robust encoding reliability. The solution enables the combinatorial expansion of coding capacity with structural layers, yielding 210, 840, and 2520 unique codes for three-, four-, and five-layer configurations, respectively. These findings validate a scalable, hull-integrated acoustic identification system that overcomes material constraints while providing high-capacity encoding for compact AUVs, significantly advancing underwater acoustic tagging technologies through physics-driven design and systematic performance validation. Full article

Deep learning network models are crucial in processing images acquired from optical, laser, and acoustic sensors in ocean intelligent perception and target detection. This work comprehensively reviews ocean intelligent perception and image processing technology, including ocean intelligent perception devices and image acquisition, image recognition and detection models, adaptive image processing processes, and coping methods for nonlinear noise interference. As the core tasks of ocean image processing, image recognition and detection network models are the research focus of this article. The focus is on the development of deep-learning network models for ocean image recognition and detection, such as SSD, R-CNN series, and YOLO series. The detailed analysis of the mathematical structure of the YOLO model and the differences between various versions, which determine the detection accuracy and inference speed, provides a deeper understanding. It also reviewed adaptive image processing processes and their critical support for ocean image recognition and detection, such as image annotation, feature enhancement, and image segmentation. Research and practical applications show that nonlinear noise significantly affects underwater image processing. When combined with image enhancement, data augmentation, and transfer learning methods, deep learning algorithms can be applied to effectively address the challenges of underwater image degradation and nonlinear noise interference. This work offers a unique perspective, highlighting the mathematical structure of the network model for ocean intelligent perception and image processing. It also discusses the benefits of DL-based denoising methods in signal–noise separation and noise suppression. With this unique perspective, this work is expected to inspire and motivate more valuable research in related fields. Full article

Underwater acoustic target recognition (UATR) technology plays a significant role in marine exploration, resource development, and national defense security. To address the limitations of existing methods in computational efficiency and recognition performance, this paper proposes an improved WS-ViT model based on Vision Transformers (ViTs). By introducing the Wavelet Transform Convolution (WTConv) module and the Simplified Linear Attention (SLAttention) module, WS-ViT can effectively extract spatiotemporal complex features, enhance classification accuracy, and significantly reduce computational costs. The model is validated using the ShipsEar dataset, and the results demonstrate that WS-ViT significantly outperforms ResNet18, VGG16, and the classical ViT model in classification accuracy, with improvements of 7.3%, 4.9%, and 2.1%, respectively. Additionally, its training efficiency is improved by 28.4% compared to ViT. This study demonstrates that WS-ViT not only enhances UATR performance but also maintains computational efficiency, providing an innovative solution for efficient and accurate underwater acoustic signal processing. Full article

Many studies have used various time-frequency feature extraction methods to convert ship-radiated noise into three-dimensional (3D) data suitable for computer vision (CV) models, which have shown good results in public datasets. However, traditional feature engineering (FE) has been enhanced to interface matching–feature engineering (IM-FE). This approach requires considerable effort in feature design, larger sample duration, or a higher upper limit of frequency. In this context, this paper proposes a one-dimensional network design for underwater acoustic target recognition (UATR-ND1D), only combined with fast Fourier transform (FFT), which can effectively alleviate the problem of IM-FE. This method is abbreviated as FFT-UATR-ND1D. FFT-UATR-ND1D was applied to the design of a one-dimensional network, named ResNet1D. Experiments were conducted on two mainstream datasets, using ResNet1D in 4320 and 360 tests, respectively. The lightweight model ResNet1D_S, with only 0.17 M parameters and 3.4 M floating point operations (FLOPs), achieved average accuracies were 97.2% and 95.20%. The larger model, ResNet1D_B, with 2.1 M parameters and 5.0 M FLOPs, both reached optimal accuracies, 98.81% and 98.42%, respectively. Compared to existing methods, those with similar parameter sizes performed 3–5% worse than the methods proposed in this paper. Additionally, methods achieving similar recognition rates require more parameters of 1 to 2 orders of magnitude and FLOPs. Full article

Cetaceans are highly adapted to the underwater environment, which is very different from the terrestrial environment. For cetaceans with neither high visual acuity nor color vision, contrast may be an important cue for visual object recognition, even in the underwater environment. Contrast is defined as the difference in luminance between an object and its background and is known to be perceived as enhanced by the luminance contrast illusion in humans. The aim of this study was to experimentally investigate whether the enhancement of contrast by the luminance contrast illusion could be observed in killer whales. Luminance discrimination tasks were performed on two captive killer whales, which were required to compare the luminance of two targets presented in monitors through an underwater window and to choose the brighter one. After baseline training, in which the target areas were surrounded by black or white inducer areas, the test condition of gray inducer areas was added. Although there were some individual differences, both individuals showed higher correct response rates for gray inducer conditions than for black and white. The results suggest that contrast was perceived as enhanced by the illusion also in killer whales and may help them to extract the contours of objects. Full article

When utilizing underwater gliders to observe submerged targets, ensuring the quality and reliability of the acquired target characteristic signals is paramount. However, the signal acquisition process is significantly compromised by noise generated from various motors on the platform, which severely contaminates the authentic target signal characteristics, thereby complicating subsequent research efforts such as target identification. Given the limited capability of wavelet transforms in processing complex non-stationary signals, and considering the non-stationary and non-linear nature of the signals in question, this study focuses on the denoising of hydroacoustic signals and the characteristics of motor noise. Building upon the traditional CEEMDAN-SVD approach, we propose an adaptive noise reduction method that combines the maximum singular value of motor noise with the differential spectrum of singular values. In particular, this paper delves into the symmetry between the noise subspace and the signal subspace in SVD decomposition. By analyzing the symmetric characteristics of their singular value distributions, the process of separating noise from signals is further optimized. The effectiveness of this denoising method is analyzed and validated through simulations and experiments. The results demonstrate that under a signal-to-noise ratio (SNR) of 3 dB, the improved CEEMDAN-SVD method reduces the mean square error by an average of 22.8% and decreases the absolute value of skewness by 27.8% compared to the traditional CEEMDAN-SVD method. These findings indicate that our proposed method exhibits superior noise reduction capabilities under strong non-stationary motor noise interference, effectively enhancing the SNR and reinforcing signal characteristics. This provides a robust foundation for improving the recognition rate of hydroacoustic targets in subsequent research. Full article