Due to the strong speckle noise caused by the seabed reverberation which makes it difficult to extract discriminating and noiseless features of a target, recognition and classification of underwater targets using side-scan sonar (SSS) images is a big challenge. Moreover, unlike classification of optical images which can use a large dataset to train the classifier, classification of SSS images usually has to exploit a very small dataset for training, which may cause classifier overfitting. Compared with traditional feature extraction methods using descriptors—such as Haar, SIFT, and LBP—deep learning-based methods are more powerful in capturing discriminating features. After training on a large optical dataset, e.g., ImageNet, direct fine-tuning method brings improvement to the sonar image classification using a small-size SSS image dataset. However, due to the different statistical characteristics between optical images and sonar images, transfer learning methods—e.g., fine-tuning—lack cross-domain adaptability, and therefore cannot achieve very satisfactory results. In this paper, a multi-domain collaborative transfer learning (MDCTL) method with multi-scale repeated attention mechanism (MSRAM) is proposed for improving the accuracy of underwater sonar image classification. In the MDCTL method, low-level characteristic similarity between SSS images and synthetic aperture radar (SAR) images, and high-level representation similarity between SSS images and optical images are used together to enhance the feature extraction ability of the deep learning model. Using different characteristics of multi-domain data to efficiently capture useful features for the sonar image classification, MDCTL offers a new way for transfer learning. MSRAM is used to effectively combine multi-scale features to make the proposed model pay more attention to the shape details of the target excluding the noise. Experimental results of classification show that, in using multi-domain data sets, the proposed method is more stable with an overall accuracy of 99.21%, bringing an improvement of 4.54% compared with the fine-tuned VGG19. Results given by diverse visualization methods also demonstrate that the method is more powerful in feature representation by using the MDCTL and MSRAM.

Due to the small sample size of underwater acoustic data and the strong noise interference caused by seabed reverberation, recognizing underwater targets in Side-Scan Sonar (SSS) images is challenging. Using a transfer-learning-based recognition method to train the backbone network on a large optical dataset (ImageNet) and fine-tuning the head network with a small SSS image dataset can improve the classification of sonar images. However, optical and sonar images have different statistical characteristics, directly affecting transfer-learning-based target recognition. In order to improve the accuracy of underwater sonar image classification, a style transformation method between optical and SSS images is proposed in this study. In the proposed method, objects with the SSS style were synthesized through content image feature extraction and image style transfer to reduce the variability of different data sources. A staged optimization strategy using multi-modal data effectively captures the anti-noise features of sonar images, providing a new learning method for transfer learning. The results of the classification experiment showed that the approach is more stable when using synthetic data and other multi-modal datasets, with an overall accuracy of 100%.

Due to the influence of environmental noise, sidelobe data, and tunnel emission under the background of multi-background underwater surveying and mapping, it is challenging to detect seafloor terrain in the background noise. Constant false alarm detection of seafloor terrain under the condition of constant false alarm probability has been an important research field. The constant false alarm detection can eliminate noise interference in a water body in the seabed topography mapping process and provide clear and accurate seabed topography information. Therefore, it is a challenging task to increase the detection probability, reduce the missing probability, and increase the detection speed in constant false alarm detection methods. Aiming at the shortcomings of the existing algorithms, this paper proposes an efficient weighted cell averaged constant false alarm detection method (WCA-CFAR). First, the cross-window reference unit sampling method is used to improve the detection speed and accurately sample the background noise unit. Then, the reference unit weighted average constant false alarm detection method is employed to calculate the detection threshold to achieve the purpose of target detection. The proposed method is verified by the simulation data detection test and a test on the actual lake test data. The test results show that the proposed method can effectively reduce the missing detection probability and improve the detection probability.

The Kirchhoff-Ray-Mode (KRM) model has been used in fishery acoustics for more than two decades to predict backscatter by individuals or aggregations of swimbladdered fish. Backscatter contributions from a fluid-like fish body and gas-filled swimbladder are coherently summed to estimate backscatter from the whole fish. For field applications, where morphologies and orientations of individual fish and their swimbladders are unknown, the coherent KRM model based on a limited number of X-ray images containing precise morphological information on swimbladders, body sizes and orientations may not provide statistically stable and robust backscatter predictions. An incoherent KRM model is proposed, which is independent of the exact knowledge of body–swimbladder acoustic interactions. When backscatter contributions from fish bodies and swimbladders are comparable, differences in coherent and incoherent KRM model target strength (TS) predictions can be as high as a few decibels. Incoherent TS predictions as a function of frequency and orientation are statistically more stable and less sensitive to variations in anatomy among individual fish than those predicted using the traditional coherent KRM model.

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.

Side-scan sonar (SSS) target recognition is an important part of building an underwater detection system and ensuring a high-precision perception of underwater information. In this paper, a novel multi-channel multi-location attention mechanism is proposed for a multi-modal phased transfer side-scan sonar target recognition model. Optical images from the ImageNet database, synthetic aperture radar (SAR) images and SSS images are used as the training datasets. The backbone network for feature extraction is transferred and learned by a staged transfer learning method. The head network used to predict the type of target extracts the attention features of SSS through a multi-channel and multi-position attention mechanism, and subsequently performs target recognition. The proposed model is tested on the SSS test dataset and evaluated using several metrics, and compared with different recognition algorithms as well. The results show that the model has better recognition accuracy and robustness for SSS targets.