In order to improve the detection performance of magnetic anomaly signal with low signal-to-noise ratio (SNR), we develop an effective method using one-dimensional convolutional neural network (1D CNN) model with multi-feature fusion. In the method, the magnetic signal is processed by Hilbert-Huang transform and discrete wavelet transform to obtain its information as pre-feature in different dimensions. The 1D CNN model with three processing blocks is used to further extract features from pre-features and identified whether the anomaly signal exists or not based on multi-feature fusion. To train the model, the positive sample set is generated by simulated signals and the measured magnetic noise, while the negative sample set is only the measured magnetic noise. Simulation results show that the proposed method has high accuracies in training and test set. A field experiment is conducted to examine the detection performance of proposed method using real data. Results show that the proposed method has good detection performances in low SNR.

Typical underwater images exhibit poor visibility because of light scattering and absorption in turbid water. To resolve the ill-posed problem, a novel underwater image enhancement method based on Removing Light Source Color and Dehazing(RLSCD) is proposed. Scene depth, which is assumed strongly correlated with attenuations, is often ignored in most of the previous methods. A new scene depth estimation, which takes the attenuations of the different light conditions into account, is first presented. Additionally, the background light is estimated based on gray open operation and scene depth to avoid that the pixels in the white objects and artificial lighting regions may be wrongly estimated as the background light. With the estimated background light and transmission map, the dehazed image is achievable. Moreover, the theoretical analysis of the Lambertian Model indicates that the disturbance of the light source color in dehazed image can be estimated, this phenomenon enables us to correct color distortion and light overcompensation by removing the disturbance. Experimental results demonstrate the RLSCD method outperforms state-of-the-art methods in terms of relatively genuine color, increased contrast, and brightness.

Parameter estimation of Direction of Arrival (DOA) using deterministic and stochastic computing paradigms is an enabling development for underwater acoustic signal processing beside its applications in the field of seismology, astronomy, earthquake and bio-medicine. In this work, the comparative study between state of the art deterministic and heuristics algorithms is presented for viable DOA estimation for different underwater dynamic objects. A Uniform Linear Array (ULA) of eight hydrophones is used for impinging acoustic waves from far-field targets. In order to evaluate the performance, the viability of innovative statistical indices is utilized to explain. Performance analysis of Genetic Algorithm(GA) and Particle Swarm Optimization(PSO) is conducted with standard counterparts including MVDR, MUSIC, ESPRIT and UESPRIT for different number of targets in terms of estimation accuracy, robustness against the number of elements and noise, cumulative distribution function of Root Mean Sqaure Error(RMSE), frequency distribution of the RMSE over the monte carlo trials, the resolution ability and computational complexity in the presence of white Gaussian measurement noise. Crammer Rao Bound (CRB) based analysis is also performed for the validation assessments and results on Monte Carlo simulations depict that the Genetic Algorithm(GA) showed the outperform counterparts on precision, convergence and complexity indices.

The salient object detection (SOD) of forward-looking sonar images plays a crucial role in underwater detection and rescue tasks. However, the existing SOD algorithms find it difficult to effectively extract salient features and spatial structure information from images with scarce semantic information, uneven intensity distribution, and high noise. Convolutional neural networks (CNNs) have strong local feature extraction capabilities, but they are easily constrained by the receptive field and lack the ability to model long-range dependencies. Transformers, with their powerful self-attention mechanism, are capable of modeling the global features of a target, but they tend to lose a significant amount of local detail. Mamba effectively models long-range dependencies in long sequence inputs through a selection mechanism, offering a novel approach to capturing long-range correlations between pixels. However, since the saliency of image pixels does not exhibit sequential dependencies, this somewhat limits Mamba’s ability to fully capture global contextual information during the forward pass. Inspired by multimodal feature fusion learning, we propose a hybrid CNN–Transformer–Mamba architecture, termed FLSSNet. FLSSNet is built upon a CNN and Transformer backbone network, integrating four core submodules to address various technical challenges: (1) The asymmetric dual encoder–decoder (ADED) is capable of simultaneously extracting features from different modalities and systematically modeling both local contextual information and global spatial structure. (2) The Transformer feature converter (TFC) module optimizes the multimodal feature fusion process through feature transformation and channel compression. (3) The long-range correlation attention (LRCA) module enhances CNN’s ability to model long-range dependencies through the collaborative use of convolutional kernels, selective sequential scanning, and attention mechanisms, while effectively suppressing noise interference. (4) The recursive contour refinement (RCR) model refines edge contour information through a layer-by-layer recursive mechanism, achieving greater precision in boundary details. The experimental results show that FLSSNet exhibits outstanding competitiveness among 25 state-of-the-art SOD methods, achieving MAE and ${\mathit{E}}_{\mathit{\xi }}$ values of 0.04 and 0.973, respectively.