Underwater Imaging Sonar Research Articles

S3Net: Semi-self-supervised neural network for visibility enhancement of speckled images

The tremendous growth of imaging sensors, which visually acquire the objects of interest, has attracted substantial interest in different engineering applications, such as underwater sonar imaging, synthetic aperture radar (SAR) imaging, and medical ultrasound imaging, etc. However, due to limited imaging environments, the intelligent vision-empowered visual information often suffers from the low usability of imaging data, e.g., Gamma-distributed speckle noise of unknown level in several imaging systems. It is commonly intractable to implement blind despeckling of visual data since the speckle noise is signal-dependent and multiplicative in nature. To improve the imaging quality, we tended to propose a semi-self-supervised neural network (termed S3Net) that enables intelligent blind despeckling in different practical applications. To be specific, our S3Net is mainly composed of two sub-networks, i.e., self-supervised feature extraction network (FExNet) and supervised feature enhancement network (FEnNet). The robustness and generalization abilities of S3Net can be accordingly guaranteed in different imaging scenarios. It is thus capable of extracting meaningful features and enhancing geometrical features under different receptive fields, leading to image quality improvement. The S3Net denoiser can balance the trade-off between noise suppression and detail preservation. Numerous experiments have been implemented on synthetic and realistic speckle-degraded imaging scenarios. Results have demonstrated that S3Net remarkably outperformed other state-of-the-art methods in terms of both qualitative and quantitative evaluations, with significant boosting on metric and visual performance.

Learning the Ego-Motion of an Underwater Imaging Sonar: A Comparative Experimental Evaluation of Novel CNN and RCNN Approaches

Learning the Ego-Motion of an Underwater Imaging Sonar: A Comparative Experimental Evaluation of Novel CNN and RCNN Approaches

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

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.

Designing a Geodesic Faceted Acoustical Volumetric Array Using a Novel Analytical Method

We present a novel analytical method as an efficient approach to design a geodesic-faceted array (GFA) for achieving a beam performance equivalent to that of a typical spherical array (SA). GFA is a triangle-based quasi-spherical configuration, which is conventionally created using the icosahedron method imitated from the geodesic dome roof construction process. In this conventional approach, the geodesic triangles have nonuniform geometries due to some distortions that occur during the random icosahedron division process. In this study, we took a paradigm shift from this approach and adopt a new technique to design a GFA that is based on uniform triangles. The characteristic equations that relate the geodesic triangle with a spherical platform were first developed as functions of the operating frequency and geometric parameters of the array. Then, the directional factor was derived to calculate the beam pattern associated with the array. A sample design of GFA for a given underwater sonar imaging system was synthesized through an optimization process. The GFA design was compared with that of a typical SA, and a reduction of 16.5% in the number of array elements was recorded in the GFA at a nearly equivalent performance. Both arrays were modeled, simulated, and analyzed using the finite element method (FEM) to validate the theoretical designs. Comparison of the results showed a high degree of compliance between the FEM and the theoretical method for both arrays. The proposed novel approach is faster and requires fewer computer resources than the FEM. Moreover, this approach is more flexible than the traditional icosahedron method in adjusting geometrical parameters in response to desired performance outputs.

Dual-Path Residual "Shrinkage" Network for Side-Scan Sonar Image Classification.

The underwater environment is complicated and changeable and contains many noises, making it difficult to detect a particular object in the underwater environment. At present, the main seabed detection technology explores the seabed environment with sonar equipment. However, the characteristics of underwater sonar imaging (e.g., low contrast, blurred edges, poor texture, and unsatisfactory quality) have serious negative influences on such image classification. Therefore, in this study, we propose a dual-path deep residual “shrinkage” network (DP-DRSN) module, which is a simple and effective neural network attention module that can classify side-scan sonar images. Specifically, the module can extract background and feature texture information of the input feature mapping through different scales (e.g., global average pooling and global max pooling), whereas scale information passes through a two-layer 1 × 1 convolution to increase nonlinearity. This helps realize cross-channel information interaction and information integration simultaneously before outputting threshold parameters in a sigmoid layer. The parameters are then multiplied by the average value of the input feature mapping to obtain a threshold, which is used to denoise the image features using the soft threshold function. The proposed DP-DRSN study provided higher classification accuracy and efficiency than other models. In this way, the feasibility and effectiveness of DP-DRSN in image classification of side-scan sonar are proven.

Research on the module of color image sonar system based on imx6ull

Underwater sonar imaging system has a wide range of application prospects, and both military and civil are increasingly valued. This paper introduces the module design of a color image sonar system based on IMX6ULL and completes the imaging of the sonar system through experiments. The main frequency and sampling frequency of the system are improved and applied to the TKIS-I color image sonar digital system. High-performance A / D is used for data acquisition and storage, and RS485 is used for long-distance transmission to the host computer to realize display and imaging. It is mainly composed of a core board, PWM signal module, communication module, sampling module, and motor drive module. The processing speed and imaging effect of the improved system are significantly improved.

Compressive Underwater Sonar Imaging with Synthetic Aperture Processing

Synthetic aperture sonar (SAS) is a technique that acquires an underwater image by synthesizing the signal received by the sonar as it moves. By forming a synthetic aperture, the sonar overcomes physical limitations and shows superior resolution when compared with use of a side-scan sonar, which is another technique for obtaining underwater images. Conventional SAS algorithms require a high concentration of sampling in the time and space domains according to Nyquist theory. Because conventional SAS algorithms go through matched filtering, side lobes are generated, resulting in deterioration of imaging performance. To overcome the shortcomings of conventional SAS algorithms, such as the low imaging performance and the requirement for high-level sampling, this paper proposes SAS algorithms applying compressive sensing (CS). SAS imaging algorithms applying CS were formulated for a single sensor and uniform line array and were verified through simulation and experimental data. The simulation showed better resolution than the ω-k algorithms, one of the representative conventional SAS algorithms, with minimal performance degradation by side lobes. The experimental data confirmed that the proposed method is superior and robust with respect to sensor loss.

Deep Learning based Object Detection via Style-transferred Underwater Sonar Images

Compared to the flourishing researches on terrestrial optical images, deep learning in underwater imaging has not been highlighted. Although some approaches applied deep learning in their underwater imaging still no major application has been found in underwater sonar imaging. Notably, the fundamental limitation in underwater image data would be the main cause of the bottleneck. To alleviate this issue, this paper introduces a simulation-generated dataset for object detection in underwater sonar images. Specifically, this paper focuses on generating real sonarlike style-transferred synthetic sonar images for network training.

A novel GPU-based sonar simulator for real-time applications

Mainly when applied in the underwater environment, sonar simulation requires great computational effort due to the complexity of acoustic physics. Simulation of sonar operation allows evaluating algorithms and control systems without going to the real underwater environment; that reduces the costs and risks of in-field experiments. This paper tackles with the problem of real-time underwater imaging sonar simulation by using the OpenGL shading language chain on GPU. Our proposed system is able to simulate two main types of acoustic devices: mechanical scanning imaging sonars and forward-looking sonars. The underwater scenario simulation is performed based on three frameworks: (i) OpenSceneGraph reproduces the ocean visual effects, (ii) Gazebo deals with physical forces, and (iii) the Robot Construction Kit controls the sonar in underwater environments. Our system exploits the rasterization pipeline in order to simulate the sonar devices, which are simulated by means of three parameters: the pulse distance, the echo intensity and the sonar field-of-view, being all calculated over observable objects shapes in the 3D rendered scene. Sonar-intrinsic operational parameters, speckle noise and object material properties are also considered as part of the acoustic image. Our evaluation demonstrated that the proposed system is able to operate close to or faster than the real-world devices. Also, our method generates visually realistic sonar images when compared with real-world sonar images of the same scenes.

Simulation study of linear power amplifier for underwater imaging sonar applications

High frequency acoustics are preferred for underwater imaging sonar applications due to its attractive characteristics such as high spatial resolution and wide bandwidth. But it suffers severe range limitation imposed by attenuation that increases approximately as frequency-squared. Hence high frequency signals with sufficient power is required to get the desired range and resolution. For underwater imaging system usually power amplifier is an integral part of wet end system. This paper discuss the simulation study of high frequency linear power amplifier for under water imaging sonar applications.

Bio-inspired broadband sonar array prototypes and underwater experiments for two- and three-dimensional acoustic imaging applications

Underwater sonar imaging relies upon the information contained in time correlation delays across numerous channels. Designing higher resolution imaging systems currently requires increasing the array aperture to wavelength ratio (L / λ), where L is the effective array length and λ is the acoustic wavelength in the medium. This fundamental constraint leads engineers down the path of adding a significant number of channels to a sonar design, which in turn increases array complexity and cost. Our research in bio-inspired sonar has revealed an approach that circumvents this constraint by exploiting bandwidth in addition to time-delay for the angular imaging process. Presented will be the results of a 2-channel underwater prototype tested in an acoustic tank and design work toward a 3-channel prototype for extending imaging to elevation angles. This work represents ongoing efforts toward developing a compact, low-cost broadband underwater sonar for near- to mid-range imaging and classification. Future integration with an autonomous underwater vehicle will demonstrate this technology for simple obstacle detection and avoidance of complex objects. [Work supported by ONR and internal investments by NUWC Division Newport.]

Underwater sonar target imaging via compressed sensing with M sequences

Due to the low sound propagation speed, the tradeoff between high azimuth resolution and wide imaging swath has severely limited the application of sonar underwater target imaging. However, based on compressed sensing (CS) technique, it is feasible to image targets with merely one pulse and thus avoid the above tradeoff. To investigate the possible waveforms for CS-based underwater imaging, the deterministic M sequences widely used in sonar applications are introduced in this paper. By analyzing the compressive matrix constructed from M sequences, the coherence parameter and the restricted isometry property (RIP) of the matrix are derived. Also, the feasibility and advances of M sequence are demonstrated by being compared with the existing Alltop sequence in underwater CS imaging framework. Finally, the results of numerical simulations and a real experiment are provided to reveal the effectiveness of the proposed signal.

Preparation and properties of 3–1 type PZT ceramics by a self-organization method

In this work, 3–1 type PZT ceramics with unidirectionally aligned channels have been fabricated by ionotropic gelation process of alginate/PZT suspensions. The 3-1 type PZT ceramics with porosity ranging from 36.9% to 60.6% were fabricated by adjusting the initial solid loading (5–25 wt%) of slurries. By incorporating CaO into PZT matrix during the process, the relative permittivity of porous ceramics was reduced, which was beneficial for enhancing the value of hydrostatic figure of merit (HFOM). With regard to the piezoelectric and mechanical properties, due to the combination of highly ordered capillary structure and doping of CaO, 3–1 type PZT ceramics demonstrated excellent performances. As a result, the prepared samples possessed a maximal HFOM value of 7.6×10−12Pa−1 and compressive strength of 125MPa. The acoustic impedance (Z) of specimens decreased linearly with increasing porosity, making them promising candidates for application to underwater sonar or medical imaging.

Wideband underwater sonar imaging via compressed sensing with scaling effect compensation

By exploiting the sparsity of the imaging scene, compressed sensing (CS) imaging may provide an effective solution to cope with the tradeoff of high azimuth resolution and wide swath in synthetic aperture sonar imaging. However, existing CS imaging methods are based on narrowband signal model and the scaling effect on the echoes is normally omitted, which may seriously affect the ultimate image reconstruction performance. This paper establishes the wideband CS underwater sonar imaging model at first, where the scaling effect and Doppler frequency shift are jointly considered. Furthermore, two wideband CS imaging approaches are discussed in uniform scale-range space and velocity-range space, respectively. Besides, modified l 1-norm minimization algorithms are proposed for the scene reconstruction in velocity-range space. Finally, numerical experiments are provided to demonstrate the effectiveness of the proposed model and the reconstruction algorithms.

FPGA-Based Adaptive Speckle Suppression Filter for Underwater Imaging Sonar

Underwater forward-looking imaging sonar (FLS) is widely used on stationary and moving platforms to overcome underwater visibility problems. The SONAR images are perturbed by a multiplicative noise called speckle, due to the coherent nature of the scattering phenomenon. Speckle reduction filters are necessary to optimize the images' exploitation procedures. The results of speckle filters may vary from one sensor and one wavelength to another; therefore, no generic de-speckling algorithm exists. Several studies have been carried out on speckle noise suppression on sidescan sonar, but the problem of speckle noise suppression for FLS has not yet been covered. A comparison of the most used classical speckle suppression filters as well as advanced wavelet-based ones was carried out. The Frost filter was found to be the most adequate for FLS data, but also the most computationally complex and not suitable for real-time processing. Two novel architectures for real-time and low-power field-programmable gate array (FPGA) implementation of the Frost speckle filter for underwater imaging sonar are presented. The proposed architectures have superior performance and power efficiency compared to standard software implementation.

Design for Underwater Aplanatic Straubel Acoustic Mirror

An aplanatic Fresnel lens was developed in a previous study to reduce the thickness of underwater acoustic lenses. Showing better convergence properties than aplanatic biconvex and hyperbolic Fresnel lenses, the aplanatic Fresnel lens had limitations in terms of frequency and number of waves owing to its shape. An underwater aplanatic Straubel acoustic mirror, the concept o which is based on a Straubel mirror, was designed to remove the two limitations of aplanatic Fresnel lenses. When the lens is used for underwater imaging sonar, the range resolution deteriorates because of these problems. The shape of the aplanatic Straubel mirror was designed using a numerical optimization method, and its convergence properties were evaluated in simulations. The mirror could correct spherical and coma aberrations in accordance with ray theory. In a comparison of the aplanatic Straubel mirror with an aplanatic Fresnel lens, the mirror had less coma aberration and field curvature than the lens. Additionally, the aplanatic Straubel mirror did not have the two limitations of the lens. Therefore, the aplanatic Straubel mirror showed better convergence than the aplanatic Fresnel lens.

Design for Underwater Aplanatic Straubel Acoustic Mirror

An aplanatic Fresnel lens was developed in a previous study to reduce the thickness of underwater acoustic lenses. Showing better convergence properties than aplanatic biconvex and hyperbolic Fresnel lenses, the aplanatic Fresnel lens had limitations in terms of frequency and number of waves owing to its shape. An underwater aplanatic Straubel acoustic mirror, the concept o which is based on a Straubel mirror, was designed to remove the two limitations of aplanatic Fresnel lenses. When the lens is used for underwater imaging sonar, the range resolution deteriorates because of these problems. The shape of the aplanatic Straubel mirror was designed using a numerical optimization method, and its convergence properties were evaluated in simulations. The mirror could correct spherical and coma aberrations in accordance with ray theory. In a comparison of the aplanatic Straubel mirror with an aplanatic Fresnel lens, the mirror had less coma aberration and field curvature than the lens. Additionally, the aplanatic Straubel mirror did not have the two limitations of the lens. Therefore, the aplanatic Straubel mirror showed better convergence than the aplanatic Fresnel lens.

Optimized simulated annealing algorithm for thinning and weighting large planar arrays

This paper proposes an optimized simulated annealing (SA) algorithm for thinning and weighting large planar arrays in 3D underwater sonar imaging systems. The optimized algorithm has been developed for use in designing a 2D planar array (a rectangular grid with a circular boundary) with a fixed side-lobe peak and a fixed current taper ratio under a narrow-band excitation. Four extensions of the SA algorithm and the procedure for the optimized SA algorithm are described. Two examples of planar arrays are used to assess the efficiency of the optimized method. The proposed method achieves a similar beam pattern performance with fewer active transducers and faster convergence ability than previous SA algorithms.

High-resolution three-dimensional imaging with synthetic-aperture forward-looking sonar systems

The concept of synthetic-aperture imaging is to utilize the scanning of data-acquisition transducer(s) to create a larger aperture span for resolution improvement. In the area of underwater sonar imaging, the synthetic-aperture imaging modality has been successfully applied to side-looking sonar systems to improve the cross-range resolution. Forward-looking sonar systems are designed to operate in simple range-profiling mode mainly for obstacle avoidance. Typically, the forward-looking systems are equipped with a physical transducer array, linear or circular. The image formation procedure produces an instantaneous two-dimensional range map converted from the time-delay profile of the echoes. As a result, the three-dimensional landscape is compressed into a two-dimensional range profile. Consequently, the depth information is lost, which seriously compromises the effectiveness of the system’s capability. In this paper, the synthetic-aperture technique is applied to the forward-looking systems. The backward propagation image algorithm is applied for three-dimensional image reconstruction. With motion compensation procedure, similar to that in the side-looking synthetic-aperture imaging, high-resolution three-dimensional dynamic imaging capability can be achieved. In addition to the retention of the depth information, the resolution of the image profiles is greatly improved with dynamic updating. [Research supported by the UC MICRO program and Sonatech.]

Sound Pressure Fields Focused Using Biconcave Acoustic Lens for Normal Incidence

The underwater imaging sonar system with an acoustic lens is again receiving considerable attention because it does not require a complex beam-forming circuit. The lens system used in this type of sonar was designed by the ray theory or by a hybrid method of the ray and wave theories. In this report, a basic analysis was performed by an analytical method using a wave theory and by a numerical method using the parabolic equation (PE) method, to determine the convergent characteristics of a biconcave lens. The pressure field focused by the biconcave lens was measured in a water tank. The biconcave lens used in the experiment is made of acrylic resin with a radius of 20 cm and a radius of curvature of 20 cm. Measurements was conducted in a water tank at a frequency of 500 kHz. Sound pressure fields around the focal region measured by the experiments agreed well with the calculated ones by the analytical and PE methods.