Improving Yolo5 for Real-Time Detection of Small Targets in Side Scan Sonar Images

Side scan sonar(SSS)is an important means to detect and locate seafloor targets.Autonomous underwater vehicles(AUVs)carrying SSS stay near the seafloor to obtain high-resolution images and provide the outline of the target for observers.The target feature information of an SSS image is similar to the background information,and a small target has less pixel information;therefore,accu-rately identifying and locating small targets in SSS images is challenging.We collect the SSS images of iron metal balls(with a diameter of 1m)and rocks to solve the problem of target misclassification.Thus,the dataset contains two types of targets,namely,‘ball’and‘rock’.With the aim to enable AUVs to accurately and automatically identify small underwater targets in SSS images,this study designs a multisize parallel convolution module embedded in state-of-the-art Yolo5.An attention mechanism transformer and a convolutional block attention module are also introduced to compare their contributions to small target detection accuracy.The performance of the proposed method is further evaluated by taking the lightweight networks Mobilenet3 and Shufflenet2 as the backbone network of Yolo5.This study focuses on the performance of convolutional neural networks for the detection of small targets in SSS images,while another comparison experiment is carried out using traditional HOG+SVM to highlight the neural network’s ability.This study aims to improve the detection accuracy while ensuring the model efficiency to meet the real-time working requirements of AUV target detection.

Fast Segmentation Method of Sonar Images for Jacket Installation Environment

It has remained a hard nut for years to segment sonar images of jacket installation environment,most of which are noisy images with inevitable blur after noise reduction.For the purpose of solutions to this problem,a fast segmen-tation algorithm is proposed on the basis of the gray value characteristics of sonar images.This algorithm is endowed with the advantage in no need of segmentation thresholds.To realize this goal,we follow the undermentioned steps:first,calcu-late the gray matrix of the fuzzy image background.After adjusting the gray value,the image is divided into three regions:background region,buffer region and target regions.Afterfiltering,we reset the pixels with gray value lower than 255 to binarize images and eliminate most artifacts.Finally,the remaining noise is removed by morphological processing.The simulation results of several sonar images show that the algorithm can segment the fuzzy sonar images quickly and effectively.Thus,the stable and feasible method is testified.

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,Rep DNet,a novel and effective despeckling convolutional neural network is proposed.Rep DNet 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 secondorder derivatives and smoothing functions.During inference,the branches are re-parameterized into a 3×3 convolution,enabling efficient inference without sacrificing accuracy.Rep DNet 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 Rep DNet as a well-balanced network,meeting the AUV computational constraints in terms of performance and latency.

A semantic segmentation-based underwater acoustic image transmission framework for cooperative SLAM

With the development of underwater sonar detection technology,simultaneous localization and mapping(SLAM)approach has attracted much attention in underwater navigation field in recent years.But the weak detection ability of a single vehicle limits the SLAM performance in wide areas.Thereby,cooperative SLAM using multiple vehicles has become an important research direction.The key factor of cooperative SLAM is timely and efficient sonar image transmission among underwater vehicles.However,the limited bandwidth of underwater acoustic channels contradicts a large amount of sonar image data.It is essential to compress the images before transmission.Recently,deep neural networks have great value in image compression by virtue of the powerful learning ability of neural networks,but the existing sonar image compression methods based on neural network usually focus on the pixel-level information without the semantic-level information.In this paper,we propose a novel underwater acoustic transmission scheme called UAT-SSIC that includes semantic segmentation-based sonar image compression(SSIC)framework and the joint source-channel codec,to improve the accuracy of the semantic information of the reconstructed sonar image at the receiver.The SSIC framework consists of Auto-Encoder structure-based sonar image compression network,which is measured by a semantic segmentation network's residual.Considering that sonar images have the characteristics of blurred target edges,the semantic segmentation network used a special dilated convolution neural network(DiCNN)to enhance segmentation accuracy by expanding the range of receptive fields.The joint source-channel codec with unequal error protection is proposed that adjusts the power level of the transmitted data,which deal with sonar image transmission error caused by the serious underwater acoustic channel.Experiment results demonstrate that our method preserves more semantic information,with advantages over existing methods at the same compression ratio.It also improves the error tolerance and packet loss resistance of transmission.

Landslide data mosaicking based on an airborne laser point cloud and multi-beam sonar images

Landslides are one of the most disastrous geological hazards in southwestern China.Once a landslide becomes unstable,it threatens the lives and safety of local residents.However,empirical studies on landslides have predominantly focused on landslides that occur on land.To this end,we aim to investigate ashore and underwater landslide data synchronously.This study proposes an optimized mosaicking method for ashore and underwater landslide data.This method fuses an airborne laser point cloud with multi-beam depth sounder images.Owing to their relatively high efficiency and large coverage area,airborne laser measurement systems are suitable for emergency investigations of landslides.Based on the airborne laser point cloud,the traversal of the point with the lowest elevation value in the point set can be used to perform rapid extraction of the crude channel boundaries.Further meticulous extraction of the channel boundaries is then implemented using the probability mean value optimization method.In addition,synthesis of the integrated ashore and underwater landslide data angle is realized using the spatial guide line between the channel boundaries and the underwater multibeam sonar images.A landslide located on the right bank of the middle reaches of the Yalong River is selected as a case study to demonstrate that the proposed method has higher precision thantraditional methods.The experimental results show that the mosaicking method in this study can meet the basic needs of landslide modeling and provide a basis for qualitative and quantitative analysis and stability prediction of landslides.

Sonar Image Registration and Mosaic Based on Line Detection and Triangle Matching

Image registration is an important research topic in the field of computer vision,in which the registration and mosaic of side-scan sonar images is the keypoints of underwater navigation.However,the image registration method of keypoints is not suitable for sonar images which do not have obvious feature points.Therefore,a method of sonar-image registration and mosaic based on line segment extraction and triangle matching is proposed in this paper.Firstly,in order to extract features from sonar image,the LSD method is introduced to detect line feature from images,and line segments are filtered by the principle of attention;after that,triangles are formed from line segments,an image transformation matrix can be calculated through the heuristic greedy algorithm from these triangles;finally,images are merged based on the transformation information.On the basis of practical tests,it is found that,the feature extraction method used in this paper can better describe the outline of underwater terrain,and there is no obvious stitching gap between the result of sonar images stitched.Experimental results show that the proposed method is effective than the keypoints method of the registration and mosaic of sonar images.

Low complexity MIMO sonar imaging using a virtual sparse linear array

A multiple-input multiple-output（MIMO） sonar can synthesize a large-aperture virtual uniform linear array（ULA） from a small number of physical elements. However, the large aperture is obtained at the cost of a great number of matched filters with much heavy computation load. To reduce the computation load, a MIMO sonar imaging method using a virtual sparse linear array（SLA） is proposed, which contains the offline and online processing. In the offline processing, the virtual ULA of the MIMO sonar is thinned to a virtual SLA by the simulated annealing algorithm, and matched filters corresponding to inactive virtual elements are removed. In the online processing, outputs of matched filters corresponding to active elements are collected for further multibeam processing and hence, the number of matched filters in the echo processing procedure is effectively reduced. Numerical simulations show that the proposed method can reduce the computation load effectively while obtaining a similar imaging performance as the traditional method.

Compensating for intensity loss in a large-aperture MIMO sonar imaging system

To reduce the computation burden of a large-aperture multiple-input multiple-output（MIMO） sonar imaging system,the phase-shift beamformer（PSBF） is used at the cost of bringing the intensity loss（IL）.The cause of the IL is analyzed in detail and a variable termed as IL factor is defined to quantify the loss amount.To compensate for the IL,two methods termed as intensity compensation for the PSBF（IC-PSBF） and the hybrid beamforming（HBF）,respectively,are proposed.The IC-PSBF uses previously estimated IL factors to compensate for output intensities of all PSBFs;and the HBF applies the IC-PSBF to the center beam region and the shifted-sideband beamformer（SSBF） to the side beam region,respectively.Numerical simulations demonstrate the effectiveness of the two proposed methods.

Automatic target recognition method for inverse synthetic aperture sonar imaging

To address the randomness of target aspect angle and the incompleteness of observed target in inverse synthetic aperture sonar（ISAS） imaging,a method for target recognition is proposed based on topology vector feature（TVF） of multiple highlights. Analysis of the projection relationship from 3 D space to 2 D imaging plane in ISAS indicates that the distance between two highlights in the cross-range scale calibrated image is determined by the distance between the corresponding physical scattering centers. Then, TVFs of different targets, which remain stable in various possibilities of target aspect angle, can be built. K-means clustering technique is used to effectively alleviate effect of the point missing due to incompleteness of the observed target. A nearest neighbor classifier is used to realize the target recognition. The ISAS experimental results using underwater scaled models are provided to demonstrate the effectiveness of the proposed method. A classification rate of 84.0% is reached.

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.