The conventional filtered-x least mean square (F-xLMS) algorithm based distributed active noise control (DANC) system's performance suffers in the presence of outliers and impulse like disturbances. In an attempt to reduce noise in such an environment Swish function based algorithms for DANC systems have been proposed presently. The Swish function makes use of the smoothness and unboundedness properties for faster convergence and eliminating vanishing gradient issue. The intention is to employ the smooth approximation of Softplus and the non-convex property of Geman-McClure estimator to propose a Softplus Geman-McClure function. In addition, the bounded nonlinearity of Welsch function which is insensitive to the outliers is utilized with the regularization property of Softsign formulating Softsign Welsch method. Henceforth, this paper proposes a family of robust algorithms employing the Swish diffusion strategy for filtered-x sign, filtered-x LMS, filtered-x Softplus Geman-McClure and filtered-x Softsign Welsch algorithms for DANC systems. The weight update rules are derived for the proposed algorithms and convergence analysis is also carried out. The suggested methods achieve faster convergence in comparison with existing techniques and approximately 1–5 dB improvement in noise cancellation for various noise inputs and impulsive noise interferences.
{"title":"A Family of Swish Diffusion Strategy Based Adaptive Algorithms for Distributed Active Noise Control","authors":"Rajapantula Kranthi;Vasundhara;Asutosh Kar;Mads Græsbøll Christensen","doi":"10.1109/OJSP.2024.3360860","DOIUrl":"https://doi.org/10.1109/OJSP.2024.3360860","url":null,"abstract":"The conventional filtered-x least mean square (F-xLMS) algorithm based distributed active noise control (DANC) system's performance suffers in the presence of outliers and impulse like disturbances. In an attempt to reduce noise in such an environment Swish function based algorithms for DANC systems have been proposed presently. The Swish function makes use of the smoothness and unboundedness properties for faster convergence and eliminating vanishing gradient issue. The intention is to employ the smooth approximation of Softplus and the non-convex property of Geman-McClure estimator to propose a Softplus Geman-McClure function. In addition, the bounded nonlinearity of Welsch function which is insensitive to the outliers is utilized with the regularization property of Softsign formulating Softsign Welsch method. Henceforth, this paper proposes a family of robust algorithms employing the Swish diffusion strategy for filtered-x sign, filtered-x LMS, filtered-x Softplus Geman-McClure and filtered-x Softsign Welsch algorithms for DANC systems. The weight update rules are derived for the proposed algorithms and convergence analysis is also carried out. The suggested methods achieve faster convergence in comparison with existing techniques and approximately 1–5 dB improvement in noise cancellation for various noise inputs and impulsive noise interferences.","PeriodicalId":73300,"journal":{"name":"IEEE open journal of signal processing","volume":"5 ","pages":"503-519"},"PeriodicalIF":0.0,"publicationDate":"2024-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10418455","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140291176","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-01-31DOI: 10.1109/OJSP.2024.3360914
Didem Dogan;Geert Leus
We consider the problem of recovering complex-valued block sparse signals with unknown borders. Such signals arise naturally in numerous applications. Several algorithms have been developed to solve the problem of unknown block partitions. In pattern-coupled sparse Bayesian learning (PCSBL), each coefficient involves its own hyperparameter and those of its immediate neighbors to exploit the block sparsity. Extended block sparse Bayesian learning (EBSBL) assumes the block sparse signal consists of correlated and overlapping blocks to enforce block correlations. We propose a simpler alternative to EBSBL and reveal the underlying relationship between the proposed method and a particular case of EBSBL. The proposed algorithm uses the fact that immediate neighboring sparse coefficients are correlated. The proposed model is similar to classical sparse Bayesian learning (SBL). However, unlike the diagonal correlation matrix in conventional SBL, the unknown correlation matrix has a tridiagonal structure to capture the correlation with neighbors. Due to the entanglement of the elements in the inverse tridiagonal matrix, instead of a direct closed-form solution, an approximate solution is proposed. The alternative algorithm avoids the high dictionary coherence in EBSBL, reduces the unknowns of EBSBL, and is computationally more efficient. The sparse reconstruction performance of the algorithm is evaluated with both correlated and uncorrelated block sparse coefficients. Simulation results demonstrate that the proposed algorithm outperforms PCSBL and correlation-based methods such as EBSBL in terms of reconstruction quality. The numerical results also show that the proposed correlated SBL algorithm can deal with isolated zeros and nonzeros as well as block sparse patterns.
{"title":"Correlated Sparse Bayesian Learning for Recovery of Block Sparse Signals With Unknown Borders","authors":"Didem Dogan;Geert Leus","doi":"10.1109/OJSP.2024.3360914","DOIUrl":"https://doi.org/10.1109/OJSP.2024.3360914","url":null,"abstract":"We consider the problem of recovering complex-valued block sparse signals with unknown borders. Such signals arise naturally in numerous applications. Several algorithms have been developed to solve the problem of unknown block partitions. In pattern-coupled sparse Bayesian learning (PCSBL), each coefficient involves its own hyperparameter and those of its immediate neighbors to exploit the block sparsity. Extended block sparse Bayesian learning (EBSBL) assumes the block sparse signal consists of correlated and overlapping blocks to enforce block correlations. We propose a simpler alternative to EBSBL and reveal the underlying relationship between the proposed method and a particular case of EBSBL. The proposed algorithm uses the fact that immediate neighboring sparse coefficients are correlated. The proposed model is similar to classical sparse Bayesian learning (SBL). However, unlike the diagonal correlation matrix in conventional SBL, the unknown correlation matrix has a tridiagonal structure to capture the correlation with neighbors. Due to the entanglement of the elements in the inverse tridiagonal matrix, instead of a direct closed-form solution, an approximate solution is proposed. The alternative algorithm avoids the high dictionary coherence in EBSBL, reduces the unknowns of EBSBL, and is computationally more efficient. The sparse reconstruction performance of the algorithm is evaluated with both correlated and uncorrelated block sparse coefficients. Simulation results demonstrate that the proposed algorithm outperforms PCSBL and correlation-based methods such as EBSBL in terms of reconstruction quality. The numerical results also show that the proposed correlated SBL algorithm can deal with isolated zeros and nonzeros as well as block sparse patterns.","PeriodicalId":73300,"journal":{"name":"IEEE open journal of signal processing","volume":"5 ","pages":"421-435"},"PeriodicalIF":0.0,"publicationDate":"2024-01-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10417118","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139749902","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-01-31DOI: 10.1109/OJSP.2024.3360896
Xudong Dong;Jun Zhao;Jingjing Pan;Meng Sun;Xiaofei Zhang;Peihao Dong;Yide Wang
Most of the existing direction of arrival (DOA) estimation methods in impulsive noise scenario are based on the fractional low-order moment statistics (FLOSs), such as the robust covariation-based (ROC), fractional low-order moment (FLOM), and phased fractional low-order moment (PFLOM). However, an unknown order moment parameter $p$ needs to be selected in these approaches, which inevitably increases the computational load if the optimal value of the parameter $p$ is determined by a large number of Monte Carlo experiments. To address this issue, we propose the adaptive order moment function (AOMF) and improved AOMF (IAOMF), which are applicable to the existing FLOSs-based methods and can also be extended to the case of sparse arrays. Moreover, we analyze the performance of AOMF and IAOMF, and simulation experiments verify the effectiveness of proposed methods.
{"title":"DOA Estimation With Nested Arrays in Impulsive Noise Scenario: An Adaptive Order Moment Strategy","authors":"Xudong Dong;Jun Zhao;Jingjing Pan;Meng Sun;Xiaofei Zhang;Peihao Dong;Yide Wang","doi":"10.1109/OJSP.2024.3360896","DOIUrl":"https://doi.org/10.1109/OJSP.2024.3360896","url":null,"abstract":"Most of the existing direction of arrival (DOA) estimation methods in impulsive noise scenario are based on the fractional low-order moment statistics (FLOSs), such as the robust covariation-based (ROC), fractional low-order moment (FLOM), and phased fractional low-order moment (PFLOM). However, an unknown order moment parameter \u0000<inline-formula><tex-math>$p$</tex-math></inline-formula>\u0000 needs to be selected in these approaches, which inevitably increases the computational load if the optimal value of the parameter \u0000<inline-formula><tex-math>$p$</tex-math></inline-formula>\u0000 is determined by a large number of Monte Carlo experiments. To address this issue, we propose the adaptive order moment function (AOMF) and improved AOMF (IAOMF), which are applicable to the existing FLOSs-based methods and can also be extended to the case of sparse arrays. Moreover, we analyze the performance of AOMF and IAOMF, and simulation experiments verify the effectiveness of proposed methods.","PeriodicalId":73300,"journal":{"name":"IEEE open journal of signal processing","volume":"5 ","pages":"493-502"},"PeriodicalIF":0.0,"publicationDate":"2024-01-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10417125","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139987096","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-01-31DOI: 10.1109/OJSP.2024.3360901
Yanbin Zou;Wenbo Wu;Jingna Fan;Huaping Liu
Source localization with time-of-arrival (TOA), time-difference-of-arrival (TDOA), time-delay (TD), received-signal-strength (RSS), or angle-of-arrival (AOA) measurements from several spatially distributed sensors is commonly used in practice. Existing analysis of the Cram $acute{text{e}}$ r-Rao lower bounds (CRLB) shows that a hybrid of two or more independent kinds of measurement has a lower CRLB than one individual type of measurement. This paper develops a unified constrained weighted-least squares (CWLS) algorithm for five types of hybrid localization systems: AOA and TOA (AOA/TOA), AOA and TDOA (AOA/TDOA), AOA and TD (AOA/TD), AOA and RSS (AOA/RSS), AOA, TOA, and RSS (AOA/TOA/RSS). These formulated CWLS problems only have one quadratic constraint, which can be effectively solved by the Lagrange multiplier method and root-finding algorithm. Extensive simulation results show that the proposed CWLS algorithms are superior to state-of-the-art algorithms and reach the CRLB.
{"title":"Constrained Weighted Least-Squares Algorithms for 3-D AOA-Based Hybrid Localization","authors":"Yanbin Zou;Wenbo Wu;Jingna Fan;Huaping Liu","doi":"10.1109/OJSP.2024.3360901","DOIUrl":"https://doi.org/10.1109/OJSP.2024.3360901","url":null,"abstract":"Source localization with time-of-arrival (TOA), time-difference-of-arrival (TDOA), time-delay (TD), received-signal-strength (RSS), or angle-of-arrival (AOA) measurements from several spatially distributed sensors is commonly used in practice. Existing analysis of the Cram \u0000<inline-formula><tex-math>$acute{text{e}}$</tex-math></inline-formula>\u0000 r-Rao lower bounds (CRLB) shows that a hybrid of two or more independent kinds of measurement has a lower CRLB than one individual type of measurement. This paper develops a unified constrained weighted-least squares (CWLS) algorithm for five types of hybrid localization systems: AOA and TOA (AOA/TOA), AOA and TDOA (AOA/TDOA), AOA and TD (AOA/TD), AOA and RSS (AOA/RSS), AOA, TOA, and RSS (AOA/TOA/RSS). These formulated CWLS problems only have one quadratic constraint, which can be effectively solved by the Lagrange multiplier method and root-finding algorithm. Extensive simulation results show that the proposed CWLS algorithms are superior to state-of-the-art algorithms and reach the CRLB.","PeriodicalId":73300,"journal":{"name":"IEEE open journal of signal processing","volume":"5 ","pages":"436-448"},"PeriodicalIF":0.0,"publicationDate":"2024-01-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10417139","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139749901","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The widespread usage of the Radio Frequency (RF) spectrum for wireless and mobile communication systems generated a significant spectrum scarcity. The Joint Radar-Communication System (JRCS) provides a framework to simultaneously utilize the allocated radar spectrum for sensing and communication purposes. Generally, a Successive Interference Cancellation (SIC) based receiver is applied to mitigate mutual interference in the JRCS configuration. However, this SIC receiver model introduces a communication residual component. In response to this issue, the article presents a novel measurement model based on communication residual components for various radar waveforms. The radar system's performance within the JRCS framework is then evaluated using the Fisher Information Matrix (FIM). The radar waveforms considered in this investigation are rectangular pulse, triangular pulse, Gaussian pulse, Linear Frequency Modulated (LFM) pulse, LFM-Gaussian pulse, and Non-Linear Frequency Modulated (NLFM) pulse. After that, the Kalman filter is deployed to estimate the target kinematics (range and range rate) of a single linearly moving target for different waveforms. Additionally, range and range rate estimation errors are quantified using the Root Mean Square Error (RMSE) metric. Furthermore, the Posterior Cramer-Rao Lower Bound (PCRLB) is derived to validate the estimation accuracy of various waveforms. The simulation results show that the range and range rate estimation errors are within the PCRLB limit at all time instants for all the designated waveforms. The results further reveal that the NLFM pulse waveform provides improved range and range rate error performance compared to all other waveforms.
无线和移动通信系统对射频（RF）频谱的广泛使用导致频谱严重短缺。联合雷达通信系统（JRCS）提供了一个框架，可同时利用分配的雷达频谱进行传感和通信。一般来说，在联合雷达-通信系统配置中，会使用基于连续干扰消除（SIC）的接收器来减轻相互干扰。然而，这种 SIC 接收机模型会引入通信残余部分。针对这一问题，文章提出了一种基于各种雷达波形的通信残差分量的新型测量模型。然后使用费雪信息矩阵（FIM）对 JRCS 框架内的雷达系统性能进行评估。本文研究的雷达波形包括矩形脉冲、三角脉冲、高斯脉冲、线性频率调制（LFM）脉冲、LFM-高斯脉冲和非线性频率调制（NLFM）脉冲。然后，利用卡尔曼滤波器估算不同波形下单个线性移动目标的运动特性（测距和测距速率）。此外，还使用均方根误差（RMSE）指标对测距和测距速率估计误差进行量化。此外，还得出了后验克拉默-拉奥下限（PCRLB），以验证各种波形的估计精度。仿真结果表明，所有指定波形的测距和测距率估计误差在所有时间瞬时都在 PCRLB 限制之内。结果进一步显示，与所有其他波形相比，NLFM 脉冲波形的测距和测距速率误差性能更佳。
{"title":"Optimum Waveform Selection for Target State Estimation in the Joint Radar-Communication System","authors":"Ashoka Chakravarthi Mahipathi;Bethi Pardha Pardhasaradhi;Srinath Gunnery;Pathipati Srihari;John d'Souza;Paramananda Jena","doi":"10.1109/OJSP.2024.3359997","DOIUrl":"https://doi.org/10.1109/OJSP.2024.3359997","url":null,"abstract":"The widespread usage of the Radio Frequency (RF) spectrum for wireless and mobile communication systems generated a significant spectrum scarcity. The Joint Radar-Communication System (JRCS) provides a framework to simultaneously utilize the allocated radar spectrum for sensing and communication purposes. Generally, a Successive Interference Cancellation (SIC) based receiver is applied to mitigate mutual interference in the JRCS configuration. However, this SIC receiver model introduces a communication residual component. In response to this issue, the article presents a novel measurement model based on communication residual components for various radar waveforms. The radar system's performance within the JRCS framework is then evaluated using the Fisher Information Matrix (FIM). The radar waveforms considered in this investigation are rectangular pulse, triangular pulse, Gaussian pulse, Linear Frequency Modulated (LFM) pulse, LFM-Gaussian pulse, and Non-Linear Frequency Modulated (NLFM) pulse. After that, the Kalman filter is deployed to estimate the target kinematics (range and range rate) of a single linearly moving target for different waveforms. Additionally, range and range rate estimation errors are quantified using the Root Mean Square Error (RMSE) metric. Furthermore, the Posterior Cramer-Rao Lower Bound (PCRLB) is derived to validate the estimation accuracy of various waveforms. The simulation results show that the range and range rate estimation errors are within the PCRLB limit at all time instants for all the designated waveforms. The results further reveal that the NLFM pulse waveform provides improved range and range rate error performance compared to all other waveforms.","PeriodicalId":73300,"journal":{"name":"IEEE open journal of signal processing","volume":"5 ","pages":"459-477"},"PeriodicalIF":0.0,"publicationDate":"2024-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10416352","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139987097","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-01-10DOI: 10.1109/OJSP.2023.3349115
MICHAEL T. MCCANN;Brendt Wohlberg
We present a new method for online selection of the penalty parameter for the alternating direction method of multipliers (ADMM) algorithm. ADMM is a widely used method for solving a range of optimization problems, including those that arise in signal and image processing. In its standard form, ADMM includes a scalar hyperparameter, known as the penalty parameter, which usually has to be tuned to achieve satisfactory empirical convergence. In this work, we develop a framework for analyzing the ADMM algorithm applied to a quadratic problem as an affine fixed point iteration. Using this framework, we develop a new method for automatically tuning the penalty parameter by detecting when it has become too large or small. We analyze this and several other methods with respect to their theoretical properties, i.e., robustness to problem transformations, and empirical performance on several optimization problems. Our proposed algorithm is based on a theoretical framework with clear, explicit assumptions and approximations, is theoretically covariant/invariant to problem transformations, is simple to implement, and exhibits competitive empirical performance.
{"title":"Robust and Simple ADMM Penalty Parameter Selection","authors":"MICHAEL T. MCCANN;Brendt Wohlberg","doi":"10.1109/OJSP.2023.3349115","DOIUrl":"https://doi.org/10.1109/OJSP.2023.3349115","url":null,"abstract":"We present a new method for online selection of the penalty parameter for the alternating direction method of multipliers (ADMM) algorithm. ADMM is a widely used method for solving a range of optimization problems, including those that arise in signal and image processing. In its standard form, ADMM includes a scalar hyperparameter, known as the penalty parameter, which usually has to be tuned to achieve satisfactory empirical convergence. In this work, we develop a framework for analyzing the ADMM algorithm applied to a quadratic problem as an affine fixed point iteration. Using this framework, we develop a new method for automatically tuning the penalty parameter by detecting when it has become too large or small. We analyze this and several other methods with respect to their theoretical properties, i.e., robustness to problem transformations, and empirical performance on several optimization problems. Our proposed algorithm is based on a theoretical framework with clear, explicit assumptions and approximations, is theoretically covariant/invariant to problem transformations, is simple to implement, and exhibits competitive empirical performance.","PeriodicalId":73300,"journal":{"name":"IEEE open journal of signal processing","volume":"5 ","pages":"402-420"},"PeriodicalIF":0.0,"publicationDate":"2024-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10387940","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139749914","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-01-09DOI: 10.1109/OJSP.2024.3351593
Daniel Waxman;Kurt Butler;Petar M. Djurić
We introduce Dagma-DCE , an interpretable and model-agnostic scheme for differentiable causal discovery. Current non- or over-parametric methods in differentiable causal discovery use opaque proxies of “independence” to justify the inclusion or exclusion of a causal relationship. We show theoretically and empirically that these proxies may be arbitrarily different than the actual causal strength. Juxtaposed with existing differentiable causal discovery algorithms, Dagma-DCE uses an interpretable measure of causal strength to define weighted adjacency matrices. In a number of simulated datasets, we show our method achieves state-of-the-art level performance. We additionally show that Dagma-DCE allows for principled thresholding and sparsity penalties by domain-experts. The code for our method is available open-source at https://github.com/DanWaxman/DAGMA-DCE , and can easily be adapted to arbitrary differentiable models.
{"title":"Dagma-DCE: Interpretable, Non-Parametric Differentiable Causal Discovery","authors":"Daniel Waxman;Kurt Butler;Petar M. Djurić","doi":"10.1109/OJSP.2024.3351593","DOIUrl":"10.1109/OJSP.2024.3351593","url":null,"abstract":"We introduce \u0000<sc>Dagma-DCE</small>\u0000, an interpretable and model-agnostic scheme for differentiable causal discovery. Current non- or over-parametric methods in differentiable causal discovery use opaque proxies of “independence” to justify the inclusion or exclusion of a causal relationship. We show theoretically and empirically that these proxies may be arbitrarily different than the actual causal strength. Juxtaposed with existing differentiable causal discovery algorithms, \u0000<sc>Dagma-DCE</small>\u0000 uses an interpretable measure of causal strength to define weighted adjacency matrices. In a number of simulated datasets, we show our method achieves state-of-the-art level performance. We additionally show that \u0000<sc>Dagma-DCE</small>\u0000 allows for principled thresholding and sparsity penalties by domain-experts. The code for our method is available open-source at \u0000<uri>https://github.com/DanWaxman/DAGMA-DCE</uri>\u0000, and can easily be adapted to arbitrary differentiable models.","PeriodicalId":73300,"journal":{"name":"IEEE open journal of signal processing","volume":"5 ","pages":"393-401"},"PeriodicalIF":0.0,"publicationDate":"2024-01-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10384714","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139450118","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-01-09DOI: 10.1109/OJSP.2024.3351472
Hui Lan;Enquan Zhang;Cheolkon Jung
Images captured through glass are usually contaminated by reflections, and the removal of them from images is a challenging task. Since the primary concern on photos is face, the face images with reflections annoy viewers severely. In this article, we propose a face reflection removal network using multispectral fusion of color (RGB) and near infrared (NIR) images, called FRRN. Due to the different spectral wavelengths of visible light [380 nm, 780 nm] and near infrared [780 nm, 2526 nm], NIR cameras are not sensitive to the visible light and thus NIR images are less corrupted by reflections. NIR images preserve structure information well and can guide the restoration process from reflections on the RGB images. Thus, we adopt multispectual fusion of RGB and NIR images for reflection removal from a face image. FRRN consists of one encoder model (contextual encoder model (CEM)) and two decoder models (NIR inference decoder model (NIDM) and image inference decoder model (IIDM)). CEM captures features from shallow to deep layers on the scene information while suppressing the sparse reflection component. NIDM infers NIR image to facilitate multi-scale guidance for reflection removal, while IIDM estimates the transmission layer with the guidance of NIDM. Besides, we present the reflection confidence generation module (RCGM) based on Laplacian convolution and channel attention-based residual block (CARB) to represent the reflection confidence in a region for reflection removal. To train FRRN, we construct a large-scale training dataset with face image and reflection layer (RGB and NIR images) and its corresponding test dataset using JAI AD-130 GE camera. Various experiments demonstrate that FRRN outperforms state-of-the-art methods for reflection removal in terms of visual quality and quantitative measurements.
{"title":"Face Reflection Removal Network Using Multispectral Fusion of RGB and NIR Images","authors":"Hui Lan;Enquan Zhang;Cheolkon Jung","doi":"10.1109/OJSP.2024.3351472","DOIUrl":"https://doi.org/10.1109/OJSP.2024.3351472","url":null,"abstract":"Images captured through glass are usually contaminated by reflections, and the removal of them from images is a challenging task. Since the primary concern on photos is face, the face images with reflections annoy viewers severely. In this article, we propose a face reflection removal network using multispectral fusion of color (RGB) and near infrared (NIR) images, called FRRN. Due to the different spectral wavelengths of visible light [380 nm, 780 nm] and near infrared [780 nm, 2526 nm], NIR cameras are not sensitive to the visible light and thus NIR images are less corrupted by reflections. NIR images preserve structure information well and can guide the restoration process from reflections on the RGB images. Thus, we adopt multispectual fusion of RGB and NIR images for reflection removal from a face image. FRRN consists of one encoder model (contextual encoder model (CEM)) and two decoder models (NIR inference decoder model (NIDM) and image inference decoder model (IIDM)). CEM captures features from shallow to deep layers on the scene information while suppressing the sparse reflection component. NIDM infers NIR image to facilitate multi-scale guidance for reflection removal, while IIDM estimates the transmission layer with the guidance of NIDM. Besides, we present the reflection confidence generation module (RCGM) based on Laplacian convolution and channel attention-based residual block (CARB) to represent the reflection confidence in a region for reflection removal. To train FRRN, we construct a large-scale training dataset with face image and reflection layer (RGB and NIR images) and its corresponding test dataset using JAI AD-130 GE camera. Various experiments demonstrate that FRRN outperforms state-of-the-art methods for reflection removal in terms of visual quality and quantitative measurements.","PeriodicalId":73300,"journal":{"name":"IEEE open journal of signal processing","volume":"5 ","pages":"383-392"},"PeriodicalIF":0.0,"publicationDate":"2024-01-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10384724","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139676115","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-01-01DOI: 10.1109/OJSP.2023.3349113
Shayan Gharib;Minh Tran;Diep Luong;Konstantinos Drossos;Tuomas Virtanen
Sound event detection systems are widely used in various applications such as surveillance and environmental monitoring where data is automatically collected, processed, and sent to a cloud for sound recognition. However, this process may inadvertently reveal sensitive information about users or their surroundings, hence raising privacy concerns. In this study, we propose a novel adversarial training method for learning representations of audio recordings that effectively prevents the detection of speech activity from the latent features of the recordings. The proposed method trains a model to generate invariant latent representations of speech-containing audio recordings that cannot be distinguished from non-speech recordings by a speech classifier. The novelty of our work is in the optimization algorithm, where the speech classifier's weights are regularly replaced with the weights of classifiers trained in a supervised manner. This increases the discrimination power of the speech classifier constantly during the adversarial training, motivating the model to generate latent representations in which speech is not distinguishable, even using new speech classifiers trained outside the adversarial training loop. The proposed method is evaluated against a baseline approach with no privacy measures and a prior adversarial training method, demonstrating a significant reduction in privacy violations compared to the baseline approach. Additionally, we show that the prior adversarial method is practically ineffective for this purpose.
{"title":"Adversarial Representation Learning for Robust Privacy Preservation in Audio","authors":"Shayan Gharib;Minh Tran;Diep Luong;Konstantinos Drossos;Tuomas Virtanen","doi":"10.1109/OJSP.2023.3349113","DOIUrl":"https://doi.org/10.1109/OJSP.2023.3349113","url":null,"abstract":"Sound event detection systems are widely used in various applications such as surveillance and environmental monitoring where data is automatically collected, processed, and sent to a cloud for sound recognition. However, this process may inadvertently reveal sensitive information about users or their surroundings, hence raising privacy concerns. In this study, we propose a novel adversarial training method for learning representations of audio recordings that effectively prevents the detection of speech activity from the latent features of the recordings. The proposed method trains a model to generate invariant latent representations of speech-containing audio recordings that cannot be distinguished from non-speech recordings by a speech classifier. The novelty of our work is in the optimization algorithm, where the speech classifier's weights are regularly replaced with the weights of classifiers trained in a supervised manner. This increases the discrimination power of the speech classifier constantly during the adversarial training, motivating the model to generate latent representations in which speech is not distinguishable, even using new speech classifiers trained outside the adversarial training loop. The proposed method is evaluated against a baseline approach with no privacy measures and a prior adversarial training method, demonstrating a significant reduction in privacy violations compared to the baseline approach. Additionally, we show that the prior adversarial method is practically ineffective for this purpose.","PeriodicalId":73300,"journal":{"name":"IEEE open journal of signal processing","volume":"5 ","pages":"294-302"},"PeriodicalIF":0.0,"publicationDate":"2024-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10379095","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139488162","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-01-01DOI: 10.1109/OJSP.2023.3349111
Yasaman Parhizkar;Gene Cheung;Andrew W. Eckford
It is a popular hypothesis in neuroscience that ganglion cells in the retina are activated by selectively detecting visual features in an observed scene. While ganglion cell firings can be predicted via data-trained deep neural nets, the networks remain indecipherable, thus providing little understanding of the cells' underlying operations. To extract knowledge from the cell firings, in this paper we learn an interpretable graph-based classifier from data to predict the firings of ganglion cells in response to visual stimuli. Specifically, we learn a positive semi-definite (PSD) metric matrix ${mathbf {M}}succeq 0$ that defines Mahalanobis distances between graph nodes (visual events) endowed with pre-computed feature vectors; the computed inter-node distances lead to edge weights and a combinatorial graph that is amenable to binary classification. Mathematically, we define the objective of metric matrix ${mathbf {M}}$ optimization using a graph adaptation of large margin nearest neighbor (LMNN), which is rewritten as a semi-definite programming (SDP) problem. We solve it efficiently via a fast approximation called Gershgorin disc perfect alignment (GDPA) linearization. The learned metric matrix ${mathbf {M}}$ provides interpretability: important features are identified along ${mathbf {M}}$ ’s diagonal, and their mutual relationships are inferred from off-diagonal terms. Our fast metric learning framework can be applied to other biological systems with pre-chosen features that require interpretation.
{"title":"Signal Processing in the Retina: Interpretable Graph Classifier to Predict Ganglion Cell Responses","authors":"Yasaman Parhizkar;Gene Cheung;Andrew W. Eckford","doi":"10.1109/OJSP.2023.3349111","DOIUrl":"10.1109/OJSP.2023.3349111","url":null,"abstract":"It is a popular hypothesis in neuroscience that ganglion cells in the retina are activated by selectively detecting visual features in an observed scene. While ganglion cell firings can be predicted via data-trained deep neural nets, the networks remain indecipherable, thus providing little understanding of the cells' underlying operations. To extract knowledge from the cell firings, in this paper we learn an interpretable graph-based classifier from data to predict the firings of ganglion cells in response to visual stimuli. Specifically, we learn a positive semi-definite (PSD) metric matrix \u0000<inline-formula><tex-math>${mathbf {M}}succeq 0$</tex-math></inline-formula>\u0000 that defines Mahalanobis distances between graph nodes (visual events) endowed with pre-computed feature vectors; the computed inter-node distances lead to edge weights and a combinatorial graph that is amenable to binary classification. Mathematically, we define the objective of metric matrix \u0000<inline-formula><tex-math>${mathbf {M}}$</tex-math></inline-formula>\u0000 optimization using a graph adaptation of large margin nearest neighbor (LMNN), which is rewritten as a semi-definite programming (SDP) problem. We solve it efficiently via a fast approximation called Gershgorin disc perfect alignment (GDPA) linearization. The learned metric matrix \u0000<inline-formula><tex-math>${mathbf {M}}$</tex-math></inline-formula>\u0000 provides interpretability: important features are identified along \u0000<inline-formula><tex-math>${mathbf {M}}$</tex-math></inline-formula>\u0000’s diagonal, and their mutual relationships are inferred from off-diagonal terms. Our fast metric learning framework can be applied to other biological systems with pre-chosen features that require interpretation.","PeriodicalId":73300,"journal":{"name":"IEEE open journal of signal processing","volume":"5 ","pages":"303-311"},"PeriodicalIF":0.0,"publicationDate":"2024-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10379097","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139388541","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：