Dajun Sun
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Dajun Sun has a total of 30 co-authors. Here are the top co-authors by number of shared publications:
- Cuie Zheng
Harbin Engineering University
2 shared publications Jidan Mei
Harbin Engn Univ, Natl Key Lab Underwater Acoust Technol, Harbin 150001, Peoples R China
2 shared publications- Tingting Teng
National Key Laboratory of Underwater Acoustic Technology, Harbin Engineering University, Harbin 150001, China
2 shared publications Chao Ma
Acoustic Science and Technology Laboratory, Harbin Engineering University Harbin People's Republic of China
1 shared publication- Chao Ma
Acoustic Science and Technology Laboratory, Harbin Engineering University, Harbin 150001, China, with the Key Laboratory of Marine Information Acquisition and Security (Harbin Engineering University), Ministry of Industry and Information Technology, Harbin 150001, China, and also with the College of Underwater Acoustic Engineering, Harbin Engineering University, Harbin 150001, China
1 shared publication - Dazhi Gao
Department of Marine Technology, Ocean University of China, Qingdao 266100, China
1 shared publication
Latest Publications
- Open Access
- 9 April 2026
- Published by MDPI AG in Journal of Marine Science and Engineering
- Vol. 14 (8) , 700
A Dual-Input Dense U-Net-Based Method for Line Spectrum Purification Under Interference Background
Line spectrum purification is a fundamental task in underwater detection and identification tasks. A dual-input architecture based on Dense U-net is introduced to extract clean line spectra from strong interference. The U-net model features a symmetric encoder–decoder structure that accepts two-dimensional data as both input and output. The DenseBlock, a core component of DenseNets, offers greater parameter efficiency compared to conventional convolutional layers. In this paper, standard convolutional layers inside the original U-net are replaced by DenseBlocks. This model possesses two input channels, thus allowing the time–frequency feature of the interference and that of the interference–target mixture to be fed simultaneously. With supervised learning, the model is capable of eliminating the strong interference components and background noise from the superimposed spectrum, thereby producing a purified target line spectrum. Compared to traditional interference suppression methods, this approach offers higher feature accuracy and greater signal-to-interference-and-noise ratio (SINR) gain. Moreover, the model is trainable using simulation datasets and then deployed to real-world measurements, demonstrating strong generalization capabilities—a valuable property given the limited availability of labeled samples in underwater detection tasks. Being data-driven, this method operates without requiring prior assumptions about the array configuration, and consequently exhibits greater resilience to array imperfections relative to conventional model-based interference suppression techniques. Simulation and experimental results demonstrate that the proposed method achieves an output SINR improvement of more than 8 dB under low SINR conditions and exhibits significantly better robustness to array position errors than conventional methods, verifying its excellent line spectrum purification capability.
- 9 April 2026
- Published by MDPI AG in Journal of Marine Science and Engineering
- Vol. 14 (8) , 700
Open AccessA Dual-Input Dense U-Net-Based Method for Line Spectrum Purification Under Interference Background
Line spectrum purification is a fundamental task in underwater detection and identification tasks. A dual-input architecture based on Dense U-net is introduced to extract clean line spectra from strong interference. The U-net model features a symmetric encoder–decoder structure that accepts two-dimensional data as both input and output. The DenseBlock, a core component of DenseNets, offers greater parameter efficiency compared to conventional convolutional layers. In this paper, standard convolutional layers inside the original U-net are replaced by DenseBlocks. This model possesses two input channels, thus allowing the time–frequency feature of the interference and that of the interference–target mixture to be fed simultaneously. With supervised learning, the model is capable of eliminating the strong interference components and background noise from the superimposed spectrum, thereby producing a purified target line spectrum. Compared to traditional interference suppression methods, this approach offers higher feature accuracy and greater signal-to-interference-and-noise ratio (SINR) gain. Moreover, the model is trainable using simulation datasets and then deployed to real-world measurements, demonstrating strong generalization capabilities—a valuable property given the limited availability of labeled samples in underwater detection tasks. Being data-driven, this method operates without requiring prior assumptions about the array configuration, and consequently exhibits greater resilience to array imperfections relative to conventional model-based interference suppression techniques. Simulation and experimental results demonstrate that the proposed method achieves an output SINR improvement of more than 8 dB under low SINR conditions and exhibits significantly better robustness to array position errors than conventional methods, verifying its excellent line spectrum purification capability.
- Open Access
- 1 February 2025
- Published by Acoustical Society of America (ASA) in JASA Express Letters
- Vol. 5 (2)
Sonar micro-Doppler effect induced by ship oscillatory motion in waves: Phenomenon, model, and experimentation
Traditional research attributes the modulated sidebands of ship radiation line-spectrum signals to reflections off rough ocean surfaces, but the Doppler effect caused by vessel motion may also contribute. This study establishes a micro-Doppler signal model based on wave-induced ship swaying, explaining that this phenomenon arises from the ship oscillatory motion in waves. The sea trial results demonstrate that the peak wave frequency, the frequency of ship micro-motions, and the micro-Doppler characteristic frequency recorded by the beacon exhibit a sudden frequency shift occurring synchronously from 0.075 to 0.115 Hz, thereby verifying the validity of the proposed model.
- 1 February 2025
- Published by Acoustical Society of America (ASA) in JASA Express Letters
- Vol. 5 (2)
Open AccessSonar micro-Doppler effect induced by ship oscillatory motion in waves: Phenomenon, model, and experimentation
Traditional research attributes the modulated sidebands of ship radiation line-spectrum signals to reflections off rough ocean surfaces, but the Doppler effect caused by vessel motion may also contribute. This study establishes a micro-Doppler signal model based on wave-induced ship swaying, explaining that this phenomenon arises from the ship oscillatory motion in waves. The sea trial results demonstrate that the peak wave frequency, the frequency of ship micro-motions, and the micro-Doppler characteristic frequency recorded by the beacon exhibit a sudden frequency shift occurring synchronously from 0.075 to 0.115 Hz, thereby verifying the validity of the proposed model.
- 21 May 2024
- Published by Elsevier in Sensors and Actuators A: Physical
- Vol. 375
Research on high sound pressure sensitivity and low noise hydrophones for underwater gliders
- 21 May 2024
- Published by Elsevier in Sensors and Actuators A: Physical
- Vol. 375
Research on high sound pressure sensitivity and low noise hydrophones for underwater gliders
- Open Access
- 14 April 2024
- Published by MDPI AG in Remote Sensing
- Vol. 16 (8) , 1391
A Deep Learning Localization Method for Acoustic Source via Improved Input Features and Network Structure
Shallow water passive source localization is an essential problem in underwater detection and localization. Traditional matched-field processing (MFP) methods are sensitive to environment mismatches. Many neural network localization methods still have room for improvement in accuracy if they are further adjusted to underwater acoustic characteristics. To address these problems, we propose a deep learning localization method via improved input features and network structure, which can effectively estimate the depth and the closest point of approach (CPA) range of the acoustic source. Firstly, we put forward a feature preprocessing scheme to enhance the localization accuracy and robustness. Secondly, we design a deep learning network structure to improve the localization accuracy further. Finally, we propose a method of visualizing the network to optimize the estimated localization results. Simulations show that the accuracy of the proposed method is better than other compared features and network structures, and the robustness is significantly better than that of the MFP methods. Experimental results further prove the effectiveness of the proposed method.
- 14 April 2024
- Published by MDPI AG in Remote Sensing
- Vol. 16 (8) , 1391
Open AccessA Deep Learning Localization Method for Acoustic Source via Improved Input Features and Network Structure
Shallow water passive source localization is an essential problem in underwater detection and localization. Traditional matched-field processing (MFP) methods are sensitive to environment mismatches. Many neural network localization methods still have room for improvement in accuracy if they are further adjusted to underwater acoustic characteristics. To address these problems, we propose a deep learning localization method via improved input features and network structure, which can effectively estimate the depth and the closest point of approach (CPA) range of the acoustic source. Firstly, we put forward a feature preprocessing scheme to enhance the localization accuracy and robustness. Secondly, we design a deep learning network structure to improve the localization accuracy further. Finally, we propose a method of visualizing the network to optimize the estimated localization results. Simulations show that the accuracy of the proposed method is better than other compared features and network structures, and the robustness is significantly better than that of the MFP methods. Experimental results further prove the effectiveness of the proposed method.
- Open Access
- 18 September 2020
- Published by Institution of Engineering and Technology (IET) in IET Radar, Sonar & Navigation
- Vol. 14 (11) , 1697-1707
Improved underwater acoustic imaging with non‐uniform spatial resampling RL deconvolution
Underwater acoustic imaging (UAI) can be utilised to observe the spatial distribution of a near‐field sound source. The image quality depends on the resolution and sidelobe level of conventional beamforming. The linear array‐based UAI can be considered as deconvolution of a two‐dimensional point spread function shift‐variant model. The performance of UAI can be improved via innovative deconvolution algorithms. In this study, a non‐uniform spatial resampling Richardson–Lucy (RL) fast algorithm is designed in which the amount of samples is determined by the power of the UAI output. This allows for a significant decrease in the number of samples compared to the traditional RL algorithm with similar positioning accuracy. Computer simulations and sea trials are performed to validate the effectiveness and feasibility of the proposed method.
- 18 September 2020
- Published by Institution of Engineering and Technology (IET) in IET Radar, Sonar & Navigation
- Vol. 14 (11) , 1697-1707
Open AccessImproved underwater acoustic imaging with non‐uniform spatial resampling RL deconvolution
Underwater acoustic imaging (UAI) can be utilised to observe the spatial distribution of a near‐field sound source. The image quality depends on the resolution and sidelobe level of conventional beamforming. The linear array‐based UAI can be considered as deconvolution of a two‐dimensional point spread function shift‐variant model. The performance of UAI can be improved via innovative deconvolution algorithms. In this study, a non‐uniform spatial resampling Richardson–Lucy (RL) fast algorithm is designed in which the amount of samples is determined by the power of the UAI output. This allows for a significant decrease in the number of samples compared to the traditional RL algorithm with similar positioning accuracy. Computer simulations and sea trials are performed to validate the effectiveness and feasibility of the proposed method.
- 23 March 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Oceanic Engineering
- Vol. 46 (1) , 183-194
Acoustic Robust Velocity Measurement Algorithm Based on Variational Bayes Adaptive Kalman Filter
Acoustic Doppler velocity logs or acoustic Doppler current profilers, herein referred to as Doppler sonar, are important types of oceanographic instruments, which have been extensively used over the past 30 years. The precision of Doppler sonar is an important technical specification, which is a statement of velocity random error that partially or entirely averages out over a long period of time. This article focuses on acoustic robust velocity processing algorithm based on variational Bayes adaptive Kalman (VBAK) filter, known as the real-time estimation of velocity statistical characteristics, which may be contaminated by some unknown and nonstationary noise. It aims to solve the problem that the precision of the Doppler sonar is prone to be affected by the marine environment. The theory, simulation, and experimental results are compared with the standard Kalman filter algorithm. It shows that the VBAK algorithm may obtain a reasonable estimate of velocity even in the case of steep slope over 15° and the precision is effectively improved.
- 23 March 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Oceanic Engineering
- Vol. 46 (1) , 183-194
Acoustic Robust Velocity Measurement Algorithm Based on Variational Bayes Adaptive Kalman Filter
Acoustic Doppler velocity logs or acoustic Doppler current profilers, herein referred to as Doppler sonar, are important types of oceanographic instruments, which have been extensively used over the past 30 years. The precision of Doppler sonar is an important technical specification, which is a statement of velocity random error that partially or entirely averages out over a long period of time. This article focuses on acoustic robust velocity processing algorithm based on variational Bayes adaptive Kalman (VBAK) filter, known as the real-time estimation of velocity statistical characteristics, which may be contaminated by some unknown and nonstationary noise. It aims to solve the problem that the precision of the Doppler sonar is prone to be affected by the marine environment. The theory, simulation, and experimental results are compared with the standard Kalman filter algorithm. It shows that the VBAK algorithm may obtain a reasonable estimate of velocity even in the case of steep slope over 15° and the precision is effectively improved.
- Open Access
- 1 June 2019
- Published by Acoustical Society of America (ASA) in The Journal of the Acoustical Society of America
- Vol. 145 (6) , EL483-EL487
Modal dispersion compensation receiver for the long range shallow water acoustic communications
The received low frequency signals in the long range shallow water will suffer severe attenuation and modal dispersion effect, which will distort the signal and deteriorate the communication performance. In this letter, a modal dispersion compensation receiver is proposed, which exploits the de-dispersion transform to compensate intra-mode dispersion and the rake combination to combine the multi-mode components. With the proposed receiver, the waveform expansion of high order modes is compensated, and thus more modes can be distinguished from the noise and the rake combination gain is improved, which leads to high communication accuracy. Simulation results show that the proposed receiver can achieve about 3 dB gain for the 40 km range and signal with a center frequency 400 Hz.
- 1 June 2019
- Published by Acoustical Society of America (ASA) in The Journal of the Acoustical Society of America
- Vol. 145 (6) , EL483-EL487
Open AccessModal dispersion compensation receiver for the long range shallow water acoustic communications
The received low frequency signals in the long range shallow water will suffer severe attenuation and modal dispersion effect, which will distort the signal and deteriorate the communication performance. In this letter, a modal dispersion compensation receiver is proposed, which exploits the de-dispersion transform to compensate intra-mode dispersion and the rake combination to combine the multi-mode components. With the proposed receiver, the waveform expansion of high order modes is compensated, and thus more modes can be distinguished from the noise and the rake combination gain is improved, which leads to high communication accuracy. Simulation results show that the proposed receiver can achieve about 3 dB gain for the 40 km range and signal with a center frequency 400 Hz.
- 16 May 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Oceanic Engineering
- Vol. 45 (3) , 1063-1077
Improving the Performance of a Vector Sensor Line Array by Deconvolution
Deconvolution applied to conventional beamforming (CBF) has been demonstrated for a uniform line array and circular array of hydrophones resulting in much improved performance than CBF. However, the existing Richardson–Lucy (R–L) method is limited to arrays with a shift-invariant beam pattern. An extended R–L deconvolution algorithm is proposed in this paper so that the deconvolved CBF can be applied to arrays with a shift-variant beam pattern, such as a vector sensor line array. The new extended R–L algorithm uses the predefined beam pattern from all steering angles to replace the shift of the point spread function. It is shown with both simulated and sea data that deconvolved beam power yields a narrower beamwidth, higher peak-to-sidelobe ratio, higher left/right suppression ratio, and higher directivity index or array gain than CBF. The extended R–L algorithm is shown to yield superior performance, particularly at low signal-to-noise ratios, compared with other deconvolution algorithms, and much less computation time for the same number of iterations, which are of practical importance.
- 16 May 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Oceanic Engineering
- Vol. 45 (3) , 1063-1077
Improving the Performance of a Vector Sensor Line Array by Deconvolution
Deconvolution applied to conventional beamforming (CBF) has been demonstrated for a uniform line array and circular array of hydrophones resulting in much improved performance than CBF. However, the existing Richardson–Lucy (R–L) method is limited to arrays with a shift-invariant beam pattern. An extended R–L deconvolution algorithm is proposed in this paper so that the deconvolved CBF can be applied to arrays with a shift-variant beam pattern, such as a vector sensor line array. The new extended R–L algorithm uses the predefined beam pattern from all steering angles to replace the shift of the point spread function. It is shown with both simulated and sea data that deconvolved beam power yields a narrower beamwidth, higher peak-to-sidelobe ratio, higher left/right suppression ratio, and higher directivity index or array gain than CBF. The extended R–L algorithm is shown to yield superior performance, particularly at low signal-to-noise ratios, compared with other deconvolution algorithms, and much less computation time for the same number of iterations, which are of practical importance.