Pub Date : 2025-01-23DOI: 10.1016/j.dsp.2025.105022
Tomáš Kerepecký , Filip Šroubek , Jan Flusser
We propose a novel approach to enhance image demosaicking algorithms using implicit neural representations (INR). Our method employs a multi-layer perceptron to encode RGB images, combining original Bayer measurements with an initial estimate from existing demosaicking methods to achieve superior reconstructions. A key innovation is the integration of two loss functions: a Bayer loss for fidelity to sensor data and a complementary loss that regularizes reconstruction using interpolated data from the initial estimate. This combination, along with INR's inherent ability to capture fine details, enables high-fidelity reconstructions that incorporate information from both sources. Furthermore, we demonstrate that INR can effectively correct artifacts in state-of-the-art demosaicking methods when input data diverge from the training distribution, such as in cases of noise or blur. This adaptability highlights the transformative potential of INR-based demosaicking, offering a robust solution to this challenging problem.
{"title":"Implicit neural representation for image demosaicking","authors":"Tomáš Kerepecký , Filip Šroubek , Jan Flusser","doi":"10.1016/j.dsp.2025.105022","DOIUrl":"10.1016/j.dsp.2025.105022","url":null,"abstract":"<div><div>We propose a novel approach to enhance image demosaicking algorithms using implicit neural representations (INR). Our method employs a multi-layer perceptron to encode RGB images, combining original Bayer measurements with an initial estimate from existing demosaicking methods to achieve superior reconstructions. A key innovation is the integration of two loss functions: a Bayer loss for fidelity to sensor data and a complementary loss that regularizes reconstruction using interpolated data from the initial estimate. This combination, along with INR's inherent ability to capture fine details, enables high-fidelity reconstructions that incorporate information from both sources. Furthermore, we demonstrate that INR can effectively correct artifacts in state-of-the-art demosaicking methods when input data diverge from the training distribution, such as in cases of noise or blur. This adaptability highlights the transformative potential of INR-based demosaicking, offering a robust solution to this challenging problem.</div></div>","PeriodicalId":51011,"journal":{"name":"Digital Signal Processing","volume":"159 ","pages":"Article 105022"},"PeriodicalIF":2.9,"publicationDate":"2025-01-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143143399","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-23DOI: 10.1016/j.dsp.2025.105020
Junwei Jin , Xianzheng Zhu , Yun Geng , Jiahang Liu , Yanting Li , Jing Liang , C.L. Philip Chen , Peng Li
The Broad Learning System (BLS) is recognized for its adept balance between efficiency and accuracy, displaying notable performance in image classification tasks owing to its streamlined network architecture and effective learning methodology. However, it faces significant challenges due to two prominent deficiencies that notably impede its learning efficacy. Firstly, the rigid binary labeling strategy inherent in BLS-based models imposes constraints on the model's adaptability. Additionally, the resultant broad features often exhibit redundancy, posing a risk of incorporating extraneous features. To address these issues, this article proposes three refined BLS-based models. Initially, a retargeting methodology is integrated into the standard BLS framework to alleviate constraints on regression targets, introducing the -based retargeted BLS (L2ReBLS) model. Subsequently, to mitigate the adverse effects of redundant features, the regularizer is adopted as a replacement for the Frobenius norm in feature selection, resulting in the L21ReBLS model. Furthermore, the projection matrix of BLS is concurrently constrained with and regularization method simultaneously. Efficient iterative optimization methodologies via the alternating direction method of multipliers are devised for the purpose of solving the proposed approaches. Ultimately, comprehensive experiments conducted on diverse image databases are to highlight the superior performance of our proposed approaches in comparison to other state-of-the-art classification algorithms.
{"title":"Retargeted broad learning systems for image classification","authors":"Junwei Jin , Xianzheng Zhu , Yun Geng , Jiahang Liu , Yanting Li , Jing Liang , C.L. Philip Chen , Peng Li","doi":"10.1016/j.dsp.2025.105020","DOIUrl":"10.1016/j.dsp.2025.105020","url":null,"abstract":"<div><div>The Broad Learning System (BLS) is recognized for its adept balance between efficiency and accuracy, displaying notable performance in image classification tasks owing to its streamlined network architecture and effective learning methodology. However, it faces significant challenges due to two prominent deficiencies that notably impede its learning efficacy. Firstly, the rigid binary labeling strategy inherent in BLS-based models imposes constraints on the model's adaptability. Additionally, the resultant broad features often exhibit redundancy, posing a risk of incorporating extraneous features. To address these issues, this article proposes three refined BLS-based models. Initially, a retargeting methodology is integrated into the standard BLS framework to alleviate constraints on regression targets, introducing the <span><math><msub><mrow><mi>ℓ</mi></mrow><mrow><mn>2</mn></mrow></msub></math></span>-based retargeted BLS (L2ReBLS) model. Subsequently, to mitigate the adverse effects of redundant features, the <span><math><msub><mrow><mi>ℓ</mi></mrow><mrow><mn>2</mn><mo>,</mo><mn>1</mn></mrow></msub></math></span> regularizer is adopted as a replacement for the Frobenius norm in feature selection, resulting in the L21ReBLS model. Furthermore, the projection matrix of BLS is concurrently constrained with <span><math><msub><mrow><mi>ℓ</mi></mrow><mrow><mn>2</mn></mrow></msub></math></span> and <span><math><msub><mrow><mi>ℓ</mi></mrow><mrow><mn>2</mn><mo>,</mo><mn>1</mn></mrow></msub></math></span> regularization method simultaneously. Efficient iterative optimization methodologies via the alternating direction method of multipliers are devised for the purpose of solving the proposed approaches. Ultimately, comprehensive experiments conducted on diverse image databases are to highlight the superior performance of our proposed approaches in comparison to other state-of-the-art classification algorithms.</div></div>","PeriodicalId":51011,"journal":{"name":"Digital Signal Processing","volume":"159 ","pages":"Article 105020"},"PeriodicalIF":2.9,"publicationDate":"2025-01-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143143972","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-23DOI: 10.1016/j.dsp.2025.105004
Salim Çınar , Alinda Ezgi Gerçek , Ahmet Ertuğrul Bilgiç , Özgür Özdemir
This study proposes an automated system to remove noise from photoacoustic (PA) signal using Independent Component Analysis (ICA). PPA signals suffer from optical and acoustic noise that degrades image quality due to the low intensity of laser light permissible in tissues. Our approach Catch Photoacoustic Peak - Independent Component Analysis (CPP-ICA), addresses this issue by applying smoothing and ICA to reduce noise without distorting PA signal characteristics. This ultimately enhances image quality while preserving important details. All independent components (ICs) of smoothed PA signal extracted using the FastICA method are processed based on their maximum peak regions, eliminating the need for manual selection of ICs for each dataset. This enables the noise removal system to operate automatically without requiring adjustments for different PA sources. Experimental results and comparative simulations with the Wavelet Denoising method show significant improvements in noise reduction performance. Our proposed technique improved the Contrast-to-Noise Ratio (CNR) and Signal-to-Noise Ratio (SNR) by 6 dB to 20 dB in experimental studies compared to the Wavelet Denoising approach, while preserving image details with minimal blurring.
{"title":"Automated noise removal system for photoacoustic imaging using independent component analysis","authors":"Salim Çınar , Alinda Ezgi Gerçek , Ahmet Ertuğrul Bilgiç , Özgür Özdemir","doi":"10.1016/j.dsp.2025.105004","DOIUrl":"10.1016/j.dsp.2025.105004","url":null,"abstract":"<div><div>This study proposes an automated system to remove noise from photoacoustic (PA) signal using Independent Component Analysis (ICA). PPA signals suffer from optical and acoustic noise that degrades image quality due to the low intensity of laser light permissible in tissues. Our approach Catch Photoacoustic Peak - Independent Component Analysis (CPP-ICA), addresses this issue by applying smoothing and ICA to reduce noise without distorting PA signal characteristics. This ultimately enhances image quality while preserving important details. All independent components (ICs) of smoothed PA signal extracted using the FastICA method are processed based on their maximum peak regions, eliminating the need for manual selection of ICs for each dataset. This enables the noise removal system to operate automatically without requiring adjustments for different PA sources. Experimental results and comparative simulations with the Wavelet Denoising method show significant improvements in noise reduction performance. Our proposed technique improved the Contrast-to-Noise Ratio (CNR) and Signal-to-Noise Ratio (SNR) by 6 dB to 20 dB in experimental studies compared to the Wavelet Denoising approach, while preserving image details with minimal blurring.</div></div>","PeriodicalId":51011,"journal":{"name":"Digital Signal Processing","volume":"159 ","pages":"Article 105004"},"PeriodicalIF":2.9,"publicationDate":"2025-01-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143143936","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-22DOI: 10.1016/j.dsp.2025.105016
Sayed Mahmoud Sakhaei
Steering vector errors and interference nonstationarity are two main degrading factors of adaptive beamforming methods. The former can be mitigated by using the interference-plus-noise covariance matrix (INCM), instead of sample covariance matrix. An estimate of INCM can be obtained through the integration of the Capon spectrum, which usually approximated by a summation over a uniform grid points on the angular region of interference and noise. A simple technique to overcome the nonstationarity problem is the null widening technique using covariance matrix tapers (CMT). The first purpose of this paper is to introduce a gridless method to do the integration for calculating the INCM. The method, applicable for uniform linear arrays, applies the residue theorem to calculate the integral through converting it into an integral over a closed contour. The second purpose is to modify the beamformer such that each null can separately be widened and deepened to definitely mitigate the corresponding interference. This purpose is also attainable by the residue-based representation of INCM, which separates the contribution of each interference, based on which a flexible null widening and deepening technique is introduced by applying different CMTs on different contributions and by moving the corresponding singular point toward the unit circle.
{"title":"A residue-based robust adaptive beamforming with flexible null management","authors":"Sayed Mahmoud Sakhaei","doi":"10.1016/j.dsp.2025.105016","DOIUrl":"10.1016/j.dsp.2025.105016","url":null,"abstract":"<div><div>Steering vector errors and interference nonstationarity are two main degrading factors of adaptive beamforming methods. The former can be mitigated by using the interference-plus-noise covariance matrix (INCM), instead of sample covariance matrix. An estimate of INCM can be obtained through the integration of the Capon spectrum, which usually approximated by a summation over a uniform grid points on the angular region of interference and noise. A simple technique to overcome the nonstationarity problem is the null widening technique using covariance matrix tapers (CMT). The first purpose of this paper is to introduce a gridless method to do the integration for calculating the INCM. The method, applicable for uniform linear arrays, applies the residue theorem to calculate the integral through converting it into an integral over a closed contour. The second purpose is to modify the beamformer such that each null can separately be widened and deepened to definitely mitigate the corresponding interference. This purpose is also attainable by the residue-based representation of INCM, which separates the contribution of each interference, based on which a flexible null widening and deepening technique is introduced by applying different CMTs on different contributions and by moving the corresponding singular point toward the unit circle.</div></div>","PeriodicalId":51011,"journal":{"name":"Digital Signal Processing","volume":"159 ","pages":"Article 105016"},"PeriodicalIF":2.9,"publicationDate":"2025-01-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143143392","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-22DOI: 10.1016/j.dsp.2025.105023
Yuqing Zhang , Dongliang Xie , Dawei Luo , Baosheng Sun
Multimodal sentiment analysis aims to obtain comprehensive emotional features from multiple data sources when all modalities are accessible. However, in real-world scenarios, it is often impossible for all modalities to be available all the time. This issue leads to a significant degradation in the performance of multimodal sentiment analysis. There are two major challenges in this field: accurately identifying samples near the classification boundaries is difficult, and the recognition performance varies significantly among different modality combinations. In this article, we propose an Emotional Boundaries and Emotional Intensity Aware (EBIA) model to enhance the robustness of incomplete multimodal sentiment analysis. Specifically, we design a Boundary Fuzzy Aware (BFA) module to learn the intra-class and inter-class consistency of samples, transferring full modalities integrity information to the missing modality environment, and prompting the model to focus on samples near the class boundaries. Additionally, we introduce a Weak Modality Aware (WMA) module that calculates additional predictions for each modality, guiding the model to focus on weak modality combinations. Extensive experiments and analyses conducted on three popular benchmark datasets demonstrate the effectiveness of our proposed method compared with several baseline methods.
{"title":"Emotional boundaries and intensity aware model for incomplete multimodal sentiment analysis","authors":"Yuqing Zhang , Dongliang Xie , Dawei Luo , Baosheng Sun","doi":"10.1016/j.dsp.2025.105023","DOIUrl":"10.1016/j.dsp.2025.105023","url":null,"abstract":"<div><div>Multimodal sentiment analysis aims to obtain comprehensive emotional features from multiple data sources when all modalities are accessible. However, in real-world scenarios, it is often impossible for all modalities to be available all the time. This issue leads to a significant degradation in the performance of multimodal sentiment analysis. There are two major challenges in this field: accurately identifying samples near the classification boundaries is difficult, and the recognition performance varies significantly among different modality combinations. In this article, we propose an Emotional Boundaries and Emotional Intensity Aware (EBIA) model to enhance the robustness of incomplete multimodal sentiment analysis. Specifically, we design a Boundary Fuzzy Aware (BFA) module to learn the intra-class and inter-class consistency of samples, transferring full modalities integrity information to the missing modality environment, and prompting the model to focus on samples near the class boundaries. Additionally, we introduce a Weak Modality Aware (WMA) module that calculates additional predictions for each modality, guiding the model to focus on weak modality combinations. Extensive experiments and analyses conducted on three popular benchmark datasets demonstrate the effectiveness of our proposed method compared with several baseline methods.</div></div>","PeriodicalId":51011,"journal":{"name":"Digital Signal Processing","volume":"160 ","pages":"Article 105023"},"PeriodicalIF":2.9,"publicationDate":"2025-01-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143350734","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-22DOI: 10.1016/j.dsp.2025.105019
Xiaowen Cai, Pinchun Li, Mingyuan Liu, Yangzhuo Chen, Jiajia Lu
The application of wireless communication systems is continuously increasing across various fields. However, due to complex electromagnetic interference, these systems struggle to transmit data accurately, particularly in environments where Global Navigation Satellite Systems (GNSS) are unavailable. To ensure the accuracy of positioning information, it is essential to research antijamming techniques. Interference identification serves as a prerequisite for effective antijamming strategies. This paper proposes a jamming recognition algorithm based on multistep singular spectrum analysis (MS-SSA) and the channel-spatial attention convolutional neural network (CSA-CNN). Noisy jamming signals are filtered via the MS-SSA method to enhance the characteristics of jamming signals at low jamming-to-noise ratios (JNRs). After filtering, the CSA-CNN is employed for jamming recognition, incorporating multidomain feature parameters. The CSA-CNN integrates the global attention mechanism to enhance the model's ability to address significant jamming features, thereby improving recognition performance. The experimental results indicate that MS-SSA achieves a superior filtering effect compared with conventional methods such as the wavelet and Kalman algorithms. In identifying jamming signals, the recognition accuracy of the CSA-CNN can exceed 90% at JNR=-2 dB. The CSA-CNN achieves superior recognition performance and generalizability compared to the convolutional neural network (CNN) and multi-branch CNN (MB-CNN).
{"title":"Radio jamming recognition algorithm based on MS-SSA and the CSA-CNN","authors":"Xiaowen Cai, Pinchun Li, Mingyuan Liu, Yangzhuo Chen, Jiajia Lu","doi":"10.1016/j.dsp.2025.105019","DOIUrl":"10.1016/j.dsp.2025.105019","url":null,"abstract":"<div><div>The application of wireless communication systems is continuously increasing across various fields. However, due to complex electromagnetic interference, these systems struggle to transmit data accurately, particularly in environments where Global Navigation Satellite Systems (GNSS) are unavailable. To ensure the accuracy of positioning information, it is essential to research antijamming techniques. Interference identification serves as a prerequisite for effective antijamming strategies. This paper proposes a jamming recognition algorithm based on multistep singular spectrum analysis (MS-SSA) and the channel-spatial attention convolutional neural network (CSA-CNN). Noisy jamming signals are filtered via the MS-SSA method to enhance the characteristics of jamming signals at low jamming-to-noise ratios (JNRs). After filtering, the CSA-CNN is employed for jamming recognition, incorporating multidomain feature parameters. The CSA-CNN integrates the global attention mechanism to enhance the model's ability to address significant jamming features, thereby improving recognition performance. The experimental results indicate that MS-SSA achieves a superior filtering effect compared with conventional methods such as the wavelet and Kalman algorithms. In identifying jamming signals, the recognition accuracy of the CSA-CNN can exceed 90% at JNR=-2 dB. The CSA-CNN achieves superior recognition performance and generalizability compared to the convolutional neural network (CNN) and multi-branch CNN (MB-CNN).</div></div>","PeriodicalId":51011,"journal":{"name":"Digital Signal Processing","volume":"159 ","pages":"Article 105019"},"PeriodicalIF":2.9,"publicationDate":"2025-01-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143144283","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-21DOI: 10.1016/j.dsp.2025.104993
Sara M. Mohamed , Mohammed H. Yacoub , Wafaa S. Sayed , Lobna A. Said , Ahmed G. Radwan
Trigonometric functions' efficient realization is essential for accurate computations in various applications, including chaotic systems that have a highly error-sensitivity nature. This paper studies the FPGA implementations of the trigonometric functions employing several methods, including LUT-based, polynomial approximation, and CORDIC. Additionally, it proposes a novel sine-based enhanced modified logistic map (EMLM) using the sine chaotification method and studies its chaotic properties using a bifurcation diagram. The trigonometric implementations are utilized to realize the sine chaotic map and the EMLM. Performance analyses are performed for the sine map and EMLM implementations, including time series, frequency distribution, histogram, and 0-1 test. The 0-1 test gives the best results for the chaotic maps that employed CORDIC realization. The FPGA resources are evaluated for all trigonometric implementations, indicating the hardware efficiency and suitability of the proposed implementations in potential applications. The chaotic maps implementations achieve a throughput of up to 0.78 Gbit/s in the polynomial-based Sine maps, demonstrating their high-performance capabilities while reaching an accurate solution.
{"title":"Efficient hardware implementations of trigonometric functions and their application to sine-based modified logistic map","authors":"Sara M. Mohamed , Mohammed H. Yacoub , Wafaa S. Sayed , Lobna A. Said , Ahmed G. Radwan","doi":"10.1016/j.dsp.2025.104993","DOIUrl":"10.1016/j.dsp.2025.104993","url":null,"abstract":"<div><div>Trigonometric functions' efficient realization is essential for accurate computations in various applications, including chaotic systems that have a highly error-sensitivity nature. This paper studies the FPGA implementations of the trigonometric functions employing several methods, including LUT-based, polynomial approximation, and CORDIC. Additionally, it proposes a novel sine-based enhanced modified logistic map (EMLM) using the sine chaotification method and studies its chaotic properties using a bifurcation diagram. The trigonometric implementations are utilized to realize the sine chaotic map and the EMLM. Performance analyses are performed for the sine map and EMLM implementations, including time series, frequency distribution, histogram, and 0-1 test. The 0-1 test gives the best results for the chaotic maps that employed CORDIC realization. The FPGA resources are evaluated for all trigonometric implementations, indicating the hardware efficiency and suitability of the proposed implementations in potential applications. The chaotic maps implementations achieve a throughput of up to 0.78 Gbit/s in the polynomial-based Sine maps, demonstrating their high-performance capabilities while reaching an accurate solution.</div></div>","PeriodicalId":51011,"journal":{"name":"Digital Signal Processing","volume":"159 ","pages":"Article 104993"},"PeriodicalIF":2.9,"publicationDate":"2025-01-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143143401","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-21DOI: 10.1016/j.dsp.2025.105024
Chunlei Liu , Hongwei Wang , Qian Zhang
This study introduces a four-stage method for parameter identification in non-homogeneous fractional-order Hammerstein MIMO systems, which solves the problems of difficulty in determining the initial values of parameters and high computational complexity. The method gradually shifts from homogeneous models to non-homogeneous models, which enhances the convergence stability of the algorithm and reduces the computational complexity. Initially, the system is simplified to a single homogeneous model, then each output subsystem is treated as homogeneous. Non-homogeneous characteristics are introduced in the third stage, and by the fourth, the entire system is considered non-homogeneous. This gradual refinement avoids the complexity of determining the initial values of the fractional order. The improved Levenberg-Marquardt algorithm, combined with the multi-innovation principle, enhances identification accuracy and global search performance. A numerical example and a PEMFC experiment verify the effectiveness and the superiority of the method.
{"title":"Parameter identification strategy for fractional-order hammerstein MIMO systems with PEMFC experimental validation","authors":"Chunlei Liu , Hongwei Wang , Qian Zhang","doi":"10.1016/j.dsp.2025.105024","DOIUrl":"10.1016/j.dsp.2025.105024","url":null,"abstract":"<div><div>This study introduces a four-stage method for parameter identification in non-homogeneous fractional-order Hammerstein MIMO systems, which solves the problems of difficulty in determining the initial values of parameters and high computational complexity. The method gradually shifts from homogeneous models to non-homogeneous models, which enhances the convergence stability of the algorithm and reduces the computational complexity. Initially, the system is simplified to a single homogeneous model, then each output subsystem is treated as homogeneous. Non-homogeneous characteristics are introduced in the third stage, and by the fourth, the entire system is considered non-homogeneous. This gradual refinement avoids the complexity of determining the initial values of the fractional order. The improved Levenberg-Marquardt algorithm, combined with the multi-innovation principle, enhances identification accuracy and global search performance. A numerical example and a PEMFC experiment verify the effectiveness and the superiority of the method.</div></div>","PeriodicalId":51011,"journal":{"name":"Digital Signal Processing","volume":"159 ","pages":"Article 105024"},"PeriodicalIF":2.9,"publicationDate":"2025-01-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143143971","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Underwater acoustic localization (UWAL) presents a significant challenge in numerous underwater applications. The speed of sound propagation, often treated as an unknown parameter, varies across different underwater environments, necessitating the joint estimation of both the source position and the sound propagation speed. In this study, we model the sound propagation speed as the sum of a hypothesized constant and a residual term, thereby formulating a new optimization problem to determine the source position and the residual speed. To address the rank-deficient issue, we employ a nullspace projection, enabling an coarse estimate of the source position through weighted least squares (WLS). To enhance accuracy, two strategies lead to two methods: the first utilizes perturbation analysis to estimate a correction that reduces error, while the second refines the coarse estimate using the maximum likelihood objective function and Taylor expansion. Performance analysis demonstrates that both proposed methods can achieve the Cramér–Rao lower bound (CRLB) in low-noise conditions. Simulations validate these analytical results and highlight the computational efficiency of the proposed methods.
{"title":"Joint source localization and propagation speed estimation using TDOA with hypothesized propagation speed","authors":"Shaohong Xu, Minghai Yang, Chengyu Li, Beichuan Tang, Yanbing Yang, Liangyin Chen, Yimao Sun","doi":"10.1016/j.dsp.2024.104934","DOIUrl":"10.1016/j.dsp.2024.104934","url":null,"abstract":"<div><div>Underwater acoustic localization (UWAL) presents a significant challenge in numerous underwater applications. The speed of sound propagation, often treated as an unknown parameter, varies across different underwater environments, necessitating the joint estimation of both the source position and the sound propagation speed. In this study, we model the sound propagation speed as the sum of a hypothesized constant and a residual term, thereby formulating a new optimization problem to determine the source position and the residual speed. To address the rank-deficient issue, we employ a nullspace projection, enabling an coarse estimate of the source position through weighted least squares (WLS). To enhance accuracy, two strategies lead to two methods: the first utilizes perturbation analysis to estimate a correction that reduces error, while the second refines the coarse estimate using the maximum likelihood objective function and Taylor expansion. Performance analysis demonstrates that both proposed methods can achieve the Cramér–Rao lower bound (CRLB) in low-noise conditions. Simulations validate these analytical results and highlight the computational efficiency of the proposed methods.</div></div>","PeriodicalId":51011,"journal":{"name":"Digital Signal Processing","volume":"159 ","pages":"Article 104934"},"PeriodicalIF":2.9,"publicationDate":"2025-01-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143143394","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-21DOI: 10.1016/j.dsp.2025.105014
Ruiqing Yan , Zongyao Yin , Cong Dai , Wengping Qi , Xiaojin Shi , Dan Hu , Dan Wu , Xianchuan Yu
Detecting low signal-to-noise ratios weak celestial source signals from large volumes of astronomical radio data is a significant and challenging task. Current mainstream approaches predominantly rely on manual processing, resulting in low efficiency. While a few deep learning-based methods have been applied, they typically utilize generic techniques, leading to suboptimal detection accuracy. To address these challenges, this paper proposes a novel method for detecting weak celestial source signals, tailored to the unique characteristics of celestial source data. The method integrates multi-task learning with pseudo-soft threshold residual denoising. Firstly, transfer learning is introduced to leverage a pre-trained model for extracting features from celestial source fringes and performing signal recognition. Multi-task learning is employed to enhance detection efficiency and reduce the false detection rate. Secondly, a novel pseudo-soft threshold function is proposed, and a corresponding pseudo-soft threshold residual denoising network is developed to automatically learn the optimal threshold and eliminate noise features. Additionally, a multi-layer fusion feature pyramid network is proposed to improve the extraction of features from weak celestial source fringes. Simulated data, generated based on the parameters of the Tianlai radio telescope observation system, is used to construct a training dataset. The performance of the proposed algorithm is evaluated using both simulated and real observational data. Experimental results demonstrate that the proposed method achieves satisfactory recognition accuracy, providing significant benefits for astronomers in detecting weak celestial source signals from extensive radio observation data. The code of this work will be available at https://github.com/YanRuiqing/MTL-PSTRD to facilitate reproducibility.
{"title":"Weak celestial source fringes detection based on multi-task learning and pseudo soft threshold residual denoising network","authors":"Ruiqing Yan , Zongyao Yin , Cong Dai , Wengping Qi , Xiaojin Shi , Dan Hu , Dan Wu , Xianchuan Yu","doi":"10.1016/j.dsp.2025.105014","DOIUrl":"10.1016/j.dsp.2025.105014","url":null,"abstract":"<div><div>Detecting low signal-to-noise ratios weak celestial source signals from large volumes of astronomical radio data is a significant and challenging task. Current mainstream approaches predominantly rely on manual processing, resulting in low efficiency. While a few deep learning-based methods have been applied, they typically utilize generic techniques, leading to suboptimal detection accuracy. To address these challenges, this paper proposes a novel method for detecting weak celestial source signals, tailored to the unique characteristics of celestial source data. The method integrates multi-task learning with pseudo-soft threshold residual denoising. Firstly, transfer learning is introduced to leverage a pre-trained model for extracting features from celestial source fringes and performing signal recognition. Multi-task learning is employed to enhance detection efficiency and reduce the false detection rate. Secondly, a novel pseudo-soft threshold function is proposed, and a corresponding pseudo-soft threshold residual denoising network is developed to automatically learn the optimal threshold and eliminate noise features. Additionally, a multi-layer fusion feature pyramid network is proposed to improve the extraction of features from weak celestial source fringes. Simulated data, generated based on the parameters of the Tianlai radio telescope observation system, is used to construct a training dataset. The performance of the proposed algorithm is evaluated using both simulated and real observational data. Experimental results demonstrate that the proposed method achieves satisfactory recognition accuracy, providing significant benefits for astronomers in detecting weak celestial source signals from extensive radio observation data. The code of this work will be available at <span><span>https://github.com/YanRuiqing/MTL-PSTRD</span><svg><path></path></svg></span> to facilitate reproducibility.</div></div>","PeriodicalId":51011,"journal":{"name":"Digital Signal Processing","volume":"159 ","pages":"Article 105014"},"PeriodicalIF":2.9,"publicationDate":"2025-01-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143143309","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号