Underwater target detection in forward-looking sonar(FLS)images is a challenging but promising endeavor.The existing neural-based methods yield notable progress but there remains room for improvement due to overlookin...Underwater target detection in forward-looking sonar(FLS)images is a challenging but promising endeavor.The existing neural-based methods yield notable progress but there remains room for improvement due to overlooking the unique characteristics of underwater environments.Considering the problems of low imaging resolution,complex background environment,and large changes in target imaging of underwater sonar images,this paper specifically designs a sonar images target detection Network based on Progressive sensitivity capture,named ProNet.It progressively captures the sensitive regions in the current image where potential effective targets may exist.Guided by this basic idea,the primary technical innovation of this paper is the introduction of a foundational module structure for constructing a sonar target detection backbone network.This structure employs a multi-subspace mixed convolution module that initially maps sonar images into different subspaces and extracts local contextual features using varying convolutional receptive fields within these heterogeneous subspaces.Subsequently,a Scale-aware aggregation module effectively aggregates the heterogeneous features extracted from different subspaces.Finally,the multi-scale attention structure further enhances the relational perception of the aggregated features.We evaluated ProNet on three FLS datasets of varying scenes,and experimental results indicate that ProNet outperforms the current state-of-the-art sonar image and general target detectors.展开更多
The disguised covert detection method that imitates whale calls has received great attention in recent years because it can solve the traditional problem of the trade-off between long-range detection and covert detect...The disguised covert detection method that imitates whale calls has received great attention in recent years because it can solve the traditional problem of the trade-off between long-range detection and covert detection.However,under strong reverberation conditions,traditional echo signal processing methods based on matched filtering will be greatly disturbed.Based on this,a disguised sonar signal waveform design is proposed based on imitating whale calls and computationally efficient anti-reverberation echo signal processing method.Firstly,this article proposed a disguised sonar signal waveform design method based on imitating whale calls.This method uses linear frequency modulation(LFM)signals to replace LFM-like segments in real whale calls,and extracts the envelope of the real whale call’s LFM-like segment to modify the LFM signal.Secondly,this article proposed an echo signal processing method of fractional Fourier transform(FrFT)based on target echo locating of synchronization signals.This method uses the synchronization signal to locate the target echo,and determines the step-size interval of the FrFT based on the information carried by the synchronization signal.Compared with the traditional FrFT,this method effectively reduces the amount of calculation and also improves the anti-reverberation ability.Finally,the excellent performance of the proposed method is verified by simulation results.展开更多
The risk of gas leakage due to geological flaws in offshore carbon capture, utilization, and storage, as well as leakage from underwater oil or gas pipelines, highlights the need for underwater gas leakage monitoring ...The risk of gas leakage due to geological flaws in offshore carbon capture, utilization, and storage, as well as leakage from underwater oil or gas pipelines, highlights the need for underwater gas leakage monitoring technology. Remotely operated vehicles(ROVs) and autonomous underwater vehicles(AUVs) are equipped with high-resolution imaging sonar systems that have broad application potential in underwater gas and target detection tasks. However, some bubble clusters are relatively weak scatterers, so detecting and distinguishing them against the seabed reverberation in forward-looking sonar images are challenging. This study uses the dual-tree complex wavelet transform to extract the image features of multibeam forward-looking sonar. Underwater gas leakages with different flows are classified by combining deep learning theory. A pool experiment is designed to simulate gas leakage, where sonar images are obtained for further processing. Results demonstrate that this method can detect and classify underwater gas leakage streams with high classification accuracy. This performance indicates that the method can detect gas leakage from multibeam forward-looking sonar images and has the potential to predict gas leakage flow.展开更多
For the constant false alarm rate(CFAR)detection problem of active sonar in complex environment and propagation channel,an automatic CFAR detection method of active sonar based on hierarchical filters was proposed.Fir...For the constant false alarm rate(CFAR)detection problem of active sonar in complex environment and propagation channel,an automatic CFAR detection method of active sonar based on hierarchical filters was proposed.Firstly,based on the spatial response characteristics of the array and the morphological characteristics of the target echo,a spatio-temporal filter was designed to remove the background noise and reverberation signal.Then,a linear and nonlinear combined spatial filter were designed to extract the energy accumulation area of the active sonar sonogram.Finally,the multi-dimensional features such as the spatial azimuth distribution of the target echo,the shape matching and the signal-to-noise ratio between the accumulation point and the adjacent background were calculated and fused to realize the target detection.Numerical simulation and sea test data show that,the probability of automatically detecting targets by this algorithm is larger than 85%in three typical cases,and this algorithm can detect targets when the background noise fluctuates greatly,showing that the algorithm is more robust.展开更多
针对主动声呐在水下环境对目标方位估计受低信噪比影响的问题,提出了一种基于分数阶傅里叶变换(Fractional Fourier Transform,FrFT)改进迭代自适应法的波达方向(Direction of Arrival,DOA)估计多波束声呐成像方法。首先对水听器收到的...针对主动声呐在水下环境对目标方位估计受低信噪比影响的问题,提出了一种基于分数阶傅里叶变换(Fractional Fourier Transform,FrFT)改进迭代自适应法的波达方向(Direction of Arrival,DOA)估计多波束声呐成像方法。首先对水听器收到的回波信号进行FrFT,通过FrFT预处理将宽带线性调频(Linear Frequency Modulation,LFM)信号变换为分数域的窄带信号,避免了交叉干扰项的影响;然后在FrFT域对LFM信号进行聚焦并对噪声进行抑制;最后在FrFT域内实现迭代自适应法,同时优化了功率谱估计方法以精确进行DOA估计。所提方法在低信噪比且不增加传感器阵元的情况下,相较于传统的DOA估计方法具有更好的估计精度与更小的均方根误差,可以显著提高成像效果。仿真结果表明,距离向的峰值旁瓣比可达到-13.364 dB,积分旁瓣比可达到-9.723 dB,方位向的峰值旁瓣比可达到-13.874 dB,积分旁瓣比可达到-10.034 dB。展开更多
针对水下无人航行器(underwater unmanned vehicle,UUV)主动声呐系统对信号处理实时性、能效比及集成度的需求,采用模块化设计以及软硬件协同设计思想,提出一种基于异构多处理器片上系统(multi-processor system on chip,MPSoC)的主动...针对水下无人航行器(underwater unmanned vehicle,UUV)主动声呐系统对信号处理实时性、能效比及集成度的需求,采用模块化设计以及软硬件协同设计思想,提出一种基于异构多处理器片上系统(multi-processor system on chip,MPSoC)的主动声呐实时信号处理算法的加速方案。首先研究适合边缘端部署的声呐信号处理算法;然后设计基于MPSoC的加速计算结构,将数字下变频、逆/快速傅里叶变换、波束形成等具有高计算复杂性的处理步骤移植到可编程逻辑端,实现显著加速;最后将目标检测等复杂度较低的步骤部署在处理器系统端,实现更高的灵活性。仿真及湖上试验结果表明,提出的方案可在数据更新周期的41%时间内完成1帧回波数据的实时处理,并可在复杂水下环境下实时有效探测运动目标。该方案在水下UUV主动声呐探测领域具有广阔的应用前景。展开更多
针对传统固定发射策略的主动声呐在水声信道中面临环境适配性不足,导致探测稳定性差的问题,本文提出一种基于多智能体强化学习的主动声呐发射波形与声源级的联合优化方法。采用多智能体协作学习方法,将发射波形优化与声源级优化解耦为...针对传统固定发射策略的主动声呐在水声信道中面临环境适配性不足,导致探测稳定性差的问题,本文提出一种基于多智能体强化学习的主动声呐发射波形与声源级的联合优化方法。采用多智能体协作学习方法,将发射波形优化与声源级优化解耦为多个智能体任务。引入奖励塑形方法,抑制多峰信道频谱引起的奖励信号噪声,提升智能体寻优能力,并避免子脉冲频点冲突。此外,使用双深度Q网络(double deep q-network),降低智能体Q值估计偏差并提升决策稳定性。在基于南海实测声速梯度重构的典型深海信道场景下进行了数值验证,结果表明:经所提算法优化后的信道适配度与回波信噪比调控准确性均优于对比算法,为构建具备环境自适应能力的智能主动声呐系统提供了一种可行的技术途径。展开更多
Resolution enhancement of active sonar can suppress the reverberation.While it also makes the envelope data distribution diverge from Rayleigh distribution to K-distribution.The stronger scattering speckles,the heavie...Resolution enhancement of active sonar can suppress the reverberation.While it also makes the envelope data distribution diverge from Rayleigh distribution to K-distribution.The stronger scattering speckles,the heavier of the K-distribution tails.The envelope amplitudes of these strong scattering speckles are usually very big.As the interfering target,the strong reverberation decreases the performances of the background power level estimation and the target detection.The fuzzy statistical normalization processing(FSNP) is introduced to suppress the strong reverberation firstly in this paper.Then how the strong reverberation and the FSNP affect the distribution of K-distributed sonar data is studied.The influence on the constant false alarm rate(CFAR) detection performance caused by the strong reverberation and the FSNP is also simulated and analyzed.Performance comparisons between the CFAR detector based on FSNP and the conventional CFAR detectors are carried out.The simulation results show that the strong reverberation can make the shape parameter of the interfering K-distributed data become smaller than that of the original K-distributed data.While the FSNP can suppress the strong reverberation,increase the shape parameter value,and improve the performance of the shape parameter estimator and the CFAR detector.展开更多
A tracking filter algorithm based on the maneuvering detection delay is presented in order to solve the fuzzy problem of target maneuver decision introduced by the measure？ment errors of active sonar. When the maneuv...A tracking filter algorithm based on the maneuvering detection delay is presented in order to solve the fuzzy problem of target maneuver decision introduced by the measure？ment errors of active sonar. When the maneuvering detection is unclear, two target moving hypotheses, the uniform and the maneuver, derived from the method of multiple hypothesis tracking, are generated to delay the final decision time. Then the hypothesis test statistics is constructed by using the residual sequence. The active sonar？s tracking ability of unknown prior information targets is improved due to the modified sequential probability ratio test and the integration of the advantages of strong tracking filter and the Kalman filter. Simulation results show that the algorithm is able to not only track the uniform targets accurately, but also track the maneuvering targets steadily. The effectiveness of the algorithm for real underwater acoustic targets is further verified by the sea trial data processing results.展开更多
基金supported in part by Youth Innovation Promotion Association,Chinese Academy of Sciences under Grant 2022022in part by South China Sea Nova project of Hainan Province under Grant NHXXRCXM202340in part by the Scientific Research Foundation Project of Hainan Acoustics Laboratory under grant ZKNZ2024001.
文摘Underwater target detection in forward-looking sonar(FLS)images is a challenging but promising endeavor.The existing neural-based methods yield notable progress but there remains room for improvement due to overlooking the unique characteristics of underwater environments.Considering the problems of low imaging resolution,complex background environment,and large changes in target imaging of underwater sonar images,this paper specifically designs a sonar images target detection Network based on Progressive sensitivity capture,named ProNet.It progressively captures the sensitive regions in the current image where potential effective targets may exist.Guided by this basic idea,the primary technical innovation of this paper is the introduction of a foundational module structure for constructing a sonar target detection backbone network.This structure employs a multi-subspace mixed convolution module that initially maps sonar images into different subspaces and extracts local contextual features using varying convolutional receptive fields within these heterogeneous subspaces.Subsequently,a Scale-aware aggregation module effectively aggregates the heterogeneous features extracted from different subspaces.Finally,the multi-scale attention structure further enhances the relational perception of the aggregated features.We evaluated ProNet on three FLS datasets of varying scenes,and experimental results indicate that ProNet outperforms the current state-of-the-art sonar image and general target detectors.
基金National Natural Science Foundations of China under Grant(Nos.61971307,61905175,51775377)National Key Research and Development Plan Project(No.2020YFB2010800)+5 种基金Fok Ying Tung Education Foundation(No.171055)China Postdoctoral Science Foundation(No.2020M680878)Guangdong Province Key Research and Development Plan Project(No.2020B0404030001)Tianjin Science and Technology Plan Project(No.20YDTPJC01660)Foreign Affairs Committee of China Aviation Development Sichuan Gas Turbine Research Institute(Nos.GJCZ-2020-0040,GJCZ-2020-0041)Science and Technology on Underwater Information and Control Laboratory under Grant(No.6142218081811)。
文摘The disguised covert detection method that imitates whale calls has received great attention in recent years because it can solve the traditional problem of the trade-off between long-range detection and covert detection.However,under strong reverberation conditions,traditional echo signal processing methods based on matched filtering will be greatly disturbed.Based on this,a disguised sonar signal waveform design is proposed based on imitating whale calls and computationally efficient anti-reverberation echo signal processing method.Firstly,this article proposed a disguised sonar signal waveform design method based on imitating whale calls.This method uses linear frequency modulation(LFM)signals to replace LFM-like segments in real whale calls,and extracts the envelope of the real whale call’s LFM-like segment to modify the LFM signal.Secondly,this article proposed an echo signal processing method of fractional Fourier transform(FrFT)based on target echo locating of synchronization signals.This method uses the synchronization signal to locate the target echo,and determines the step-size interval of the FrFT based on the information carried by the synchronization signal.Compared with the traditional FrFT,this method effectively reduces the amount of calculation and also improves the anti-reverberation ability.Finally,the excellent performance of the proposed method is verified by simulation results.
文摘The risk of gas leakage due to geological flaws in offshore carbon capture, utilization, and storage, as well as leakage from underwater oil or gas pipelines, highlights the need for underwater gas leakage monitoring technology. Remotely operated vehicles(ROVs) and autonomous underwater vehicles(AUVs) are equipped with high-resolution imaging sonar systems that have broad application potential in underwater gas and target detection tasks. However, some bubble clusters are relatively weak scatterers, so detecting and distinguishing them against the seabed reverberation in forward-looking sonar images are challenging. This study uses the dual-tree complex wavelet transform to extract the image features of multibeam forward-looking sonar. Underwater gas leakages with different flows are classified by combining deep learning theory. A pool experiment is designed to simulate gas leakage, where sonar images are obtained for further processing. Results demonstrate that this method can detect and classify underwater gas leakage streams with high classification accuracy. This performance indicates that the method can detect gas leakage from multibeam forward-looking sonar images and has the potential to predict gas leakage flow.
基金supported by the China State Shipbuilding Corporation Equipment Pre-Research Joint Fund(6141B03090103).
文摘For the constant false alarm rate(CFAR)detection problem of active sonar in complex environment and propagation channel,an automatic CFAR detection method of active sonar based on hierarchical filters was proposed.Firstly,based on the spatial response characteristics of the array and the morphological characteristics of the target echo,a spatio-temporal filter was designed to remove the background noise and reverberation signal.Then,a linear and nonlinear combined spatial filter were designed to extract the energy accumulation area of the active sonar sonogram.Finally,the multi-dimensional features such as the spatial azimuth distribution of the target echo,the shape matching and the signal-to-noise ratio between the accumulation point and the adjacent background were calculated and fused to realize the target detection.Numerical simulation and sea test data show that,the probability of automatically detecting targets by this algorithm is larger than 85%in three typical cases,and this algorithm can detect targets when the background noise fluctuates greatly,showing that the algorithm is more robust.
文摘针对水下无人航行器(underwater unmanned vehicle,UUV)主动声呐系统对信号处理实时性、能效比及集成度的需求,采用模块化设计以及软硬件协同设计思想,提出一种基于异构多处理器片上系统(multi-processor system on chip,MPSoC)的主动声呐实时信号处理算法的加速方案。首先研究适合边缘端部署的声呐信号处理算法;然后设计基于MPSoC的加速计算结构,将数字下变频、逆/快速傅里叶变换、波束形成等具有高计算复杂性的处理步骤移植到可编程逻辑端,实现显著加速;最后将目标检测等复杂度较低的步骤部署在处理器系统端,实现更高的灵活性。仿真及湖上试验结果表明,提出的方案可在数据更新周期的41%时间内完成1帧回波数据的实时处理,并可在复杂水下环境下实时有效探测运动目标。该方案在水下UUV主动声呐探测领域具有广阔的应用前景。
文摘针对传统固定发射策略的主动声呐在水声信道中面临环境适配性不足,导致探测稳定性差的问题,本文提出一种基于多智能体强化学习的主动声呐发射波形与声源级的联合优化方法。采用多智能体协作学习方法,将发射波形优化与声源级优化解耦为多个智能体任务。引入奖励塑形方法,抑制多峰信道频谱引起的奖励信号噪声,提升智能体寻优能力,并避免子脉冲频点冲突。此外,使用双深度Q网络(double deep q-network),降低智能体Q值估计偏差并提升决策稳定性。在基于南海实测声速梯度重构的典型深海信道场景下进行了数值验证,结果表明:经所提算法优化后的信道适配度与回波信噪比调控准确性均优于对比算法,为构建具备环境自适应能力的智能主动声呐系统提供了一种可行的技术途径。
基金supported by the National Natural Science Foundation of China(61431020,61471352,61671443)
文摘Resolution enhancement of active sonar can suppress the reverberation.While it also makes the envelope data distribution diverge from Rayleigh distribution to K-distribution.The stronger scattering speckles,the heavier of the K-distribution tails.The envelope amplitudes of these strong scattering speckles are usually very big.As the interfering target,the strong reverberation decreases the performances of the background power level estimation and the target detection.The fuzzy statistical normalization processing(FSNP) is introduced to suppress the strong reverberation firstly in this paper.Then how the strong reverberation and the FSNP affect the distribution of K-distributed sonar data is studied.The influence on the constant false alarm rate(CFAR) detection performance caused by the strong reverberation and the FSNP is also simulated and analyzed.Performance comparisons between the CFAR detector based on FSNP and the conventional CFAR detectors are carried out.The simulation results show that the strong reverberation can make the shape parameter of the interfering K-distributed data become smaller than that of the original K-distributed data.While the FSNP can suppress the strong reverberation,increase the shape parameter value,and improve the performance of the shape parameter estimator and the CFAR detector.
文摘A tracking filter algorithm based on the maneuvering detection delay is presented in order to solve the fuzzy problem of target maneuver decision introduced by the measure？ment errors of active sonar. When the maneuvering detection is unclear, two target moving hypotheses, the uniform and the maneuver, derived from the method of multiple hypothesis tracking, are generated to delay the final decision time. Then the hypothesis test statistics is constructed by using the residual sequence. The active sonar？s tracking ability of unknown prior information targets is improved due to the modified sequential probability ratio test and the integration of the advantages of strong tracking filter and the Kalman filter. Simulation results show that the algorithm is able to not only track the uniform targets accurately, but also track the maneuvering targets steadily. The effectiveness of the algorithm for real underwater acoustic targets is further verified by the sea trial data processing results.
