Signal Processing最新文献_第2页

Tensor block-block terms decomposition for matrix-valued imaging applications 矩阵值成像应用的张量分块项分解

IF 3.6 2区工程技术 Q2 ENGINEERING, ELECTRICAL & ELECTRONIC

Signal Processing

Pub Date : 2026-01-01 DOI: 10.1016/j.sigpro.2025.110482

Saulo Cardoso Barreto, Julien Flamant, Sebastian Miron, David Brie

Matrix-valued images appear in many applications, ranging from polarimetric remote sensing to medical imaging. Such images can be represented as 4th-order tensors, where the first two dimensions correspond to spatial variables and the last two encode the matrix feature in each pixel. To efficiently analyze, decompose, and process these images, this paper considers the block-block terms decomposition (2BTD), a versatile low-rank tensor decomposition model that extends bilinear matrix factorization to 4th-order tensors by representing the latter as the sum of outer products of low-rank matrix blocks. Low-rank assumptions allow for a significantly reduced number of parameters to be estimated and enable the enforcement of key physical constraints on matrix sources. We establish both necessary and sufficient conditions for the uniqueness of the 2BTD model. To enable the use of 2BTD in covariance matrix-valued imaging, we develop an optimization framework that allows efficient handling of non-negativity and symmetry constraints together with low-rank assumptions on matrix blocks. Numerical experiments on synthetic and real data from Diffusion Tensor Imaging (DTI) illustrate the potential of the 2BTD model in matrix-valued imaging, as well as its effectiveness in practical settings.

矩阵值图像出现在许多应用中，从偏振遥感到医学成像。这样的图像可以表示为四阶张量，其中前两个维度对应于空间变量，后两个维度编码每个像素中的矩阵特征。为了有效地分析、分解和处理这些图像，本文考虑了块项分解（2BTD），这是一种通用的低秩张量分解模型，通过将双线性矩阵分解表示为低秩矩阵块的外积和，将双线性矩阵分解扩展到4阶张量。低秩假设允许大大减少需要估计的参数数量，并使对矩阵源的关键物理约束得以实施。建立了2BTD模型唯一性的充分必要条件。为了在协方差矩阵值成像中使用2BTD，我们开发了一个优化框架，该框架允许有效处理非负性和对称约束以及矩阵块上的低秩假设。利用扩散张量成像（DTI）的合成数据和真实数据进行的数值实验表明，2BTD模型在矩阵值成像中的潜力及其在实际环境中的有效性。

{"title":"Tensor block-block terms decomposition for matrix-valued imaging applications","authors":"Saulo Cardoso Barreto, Julien Flamant, Sebastian Miron, David Brie","doi":"10.1016/j.sigpro.2025.110482","DOIUrl":"10.1016/j.sigpro.2025.110482","url":null,"abstract":"<div><div>Matrix-valued images appear in many applications, ranging from polarimetric remote sensing to medical imaging. Such images can be represented as 4th-order tensors, where the first two dimensions correspond to spatial variables and the last two encode the matrix feature in each pixel. To efficiently analyze, decompose, and process these images, this paper considers the block-block terms decomposition (2BTD), a versatile low-rank tensor decomposition model that extends bilinear matrix factorization to 4th-order tensors by representing the latter as the sum of outer products of low-rank matrix blocks. Low-rank assumptions allow for a significantly reduced number of parameters to be estimated and enable the enforcement of key physical constraints on matrix sources. We establish both necessary and sufficient conditions for the uniqueness of the 2BTD model. To enable the use of 2BTD in covariance matrix-valued imaging, we develop an optimization framework that allows efficient handling of non-negativity and symmetry constraints together with low-rank assumptions on matrix blocks. Numerical experiments on synthetic and real data from Diffusion Tensor Imaging (DTI) illustrate the potential of the 2BTD model in matrix-valued imaging, as well as its effectiveness in practical settings.</div></div>","PeriodicalId":49523,"journal":{"name":"Signal Processing","volume":"243 ","pages":"Article 110482"},"PeriodicalIF":3.6,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145927507","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Dual enhancement of low-light image based on single-channel curve estimation 基于单通道曲线估计的弱光图像双增强

IF 3.6 2区工程技术 Q2 ENGINEERING, ELECTRICAL & ELECTRONIC

Signal Processing

Pub Date : 2025-12-31 DOI: 10.1016/j.sigpro.2025.110484

Wei Hu, Manli Wang, Changsen Zhang

Recently, image enhancement technology based on deep learning has made great progress. However, existing supervised enhancement methods are not universally applicable due to the limitation of paired training sets. The enhanced images have flaws including overexposure, color distortion, and noise amplification. Through mathematical reasoning on the relationship between the Retinex model and two color spaces, this paper proposes a new unsupervised low-light enhancement method, namely a dual enhancement method based on single-channel curve estimation. We define the first-stage enhancing objective as a light curve estimation problem and design an illumination estimation module. New cubic curves and new exposure control losses are designed to avoid overexposed images. In the second-stage enhancement, we reconstruct the enhancement process by designing gradient-guided unsupervised losses to constrain the local perception restoration module, which can jointly solve image degradation problems such as noise and artifacts. Compared with other methods based on the Retinex model, our method does not require complex decomposition and reconstruction, and only requires fewer zero-reference low-light images to complete training. Lastly, we conduct comprehensive experiments to evaluate the results through subjective visual analysis and objective metric evaluation, demonstrating the superior performance of the proposed algorithm.

近年来，基于深度学习的图像增强技术取得了很大的进展。然而，现有的监督增强方法受成对训练集的限制，并不是普遍适用的。增强后的图像存在曝光过度、色彩失真和噪声放大等缺陷。本文通过对Retinex模型与两个色彩空间之间关系的数学推理，提出了一种新的无监督弱光增强方法，即基于单通道曲线估计的双增强方法。我们将第一阶段的增强目标定义为光曲线估计问题，并设计了照度估计模块。新的三次曲线和新的曝光控制损失的设计，以避免过度曝光的图像。在第二阶段增强中，我们通过设计梯度引导的无监督损失来约束局部感知恢复模块，重构增强过程，共同解决噪声和伪像等图像退化问题。与其他基于Retinex模型的方法相比，我们的方法不需要复杂的分解和重建，只需要更少的零参考低光图像就可以完成训练。最后，我们进行了综合实验，通过主观的视觉分析和客观的度量评价来评价结果，证明了所提算法的优越性能。

{"title":"Dual enhancement of low-light image based on single-channel curve estimation","authors":"Wei Hu, Manli Wang, Changsen Zhang","doi":"10.1016/j.sigpro.2025.110484","DOIUrl":"10.1016/j.sigpro.2025.110484","url":null,"abstract":"<div><div>Recently, image enhancement technology based on deep learning has made great progress. However, existing supervised enhancement methods are not universally applicable due to the limitation of paired training sets. The enhanced images have flaws including overexposure, color distortion, and noise amplification. Through mathematical reasoning on the relationship between the Retinex model and two color spaces, this paper proposes a new unsupervised low-light enhancement method, namely a dual enhancement method based on single-channel curve estimation. We define the first-stage enhancing objective as a light curve estimation problem and design an illumination estimation module. New cubic curves and new exposure control losses are designed to avoid overexposed images. In the second-stage enhancement, we reconstruct the enhancement process by designing gradient-guided unsupervised losses to constrain the local perception restoration module, which can jointly solve image degradation problems such as noise and artifacts. Compared with other methods based on the Retinex model, our method does not require complex decomposition and reconstruction, and only requires fewer zero-reference low-light images to complete training. Lastly, we conduct comprehensive experiments to evaluate the results through subjective visual analysis and objective metric evaluation, demonstrating the superior performance of the proposed algorithm.</div></div>","PeriodicalId":49523,"journal":{"name":"Signal Processing","volume":"243 ","pages":"Article 110484"},"PeriodicalIF":3.6,"publicationDate":"2025-12-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145927511","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Double low-rank 4D tensor decomposition for circular RIS-aided mmWave MIMO-NOMA system channel estimation in mobility scenarios 基于双低秩四维张量分解的环形ris辅助毫米波MIMO-NOMA系统信道估计

IF 3.6 2区工程技术 Q2 ENGINEERING, ELECTRICAL & ELECTRONIC

Signal Processing

Pub Date : 2025-12-28 DOI: 10.1016/j.sigpro.2025.110466

Wanyuan Cai , Youming Li , Menglei Sheng , Mingjun Huang , Qinke Qi , Shunli Hong

In this paper, we consider a downlink channel estimation problem for circular reconfigurable intelligent surface (RIS)-aided millimeter-wave (mmWave) multiple-input multiple-output non-orthogonal multiple access (MIMO-NOMA) system in mobility scenarios. First, we propose a subframe partitioning scheme to facilitate the modeling of the received signal as a fourth-order tensor satisfying a canonical polyadic decomposition (CPD) form, thereby formulating the channel estimation problem as tensor decomposition and parameter extraction problems. Then, by exploiting both the global and local low-rank properties of the received signal, we propose a double low-rank 4D tensor decomposition model to decompose the received signal into four factor matrices, which is efficiently solved via alternating direction method of multipliers (ADMM). Subsequently, we propose a two-stage parameter estimation method based on the Jacobi-Anger expansion and the special structure of circular RIS to uniquely decouple the angle parameters. Furthermore, the time delay, Doppler shift, and channel gain parameters can also be estimated without ambiguities, and their estimation accuracy can be efficiently improved, especially at low signal-to-noise ratio (SNR). Finally, a concise closed-form expression for the Cramér-Rao bound (CRB) is derived as a performance benchmark. Numerical experiments are conducted to demonstrate the effectiveness of the proposed method compared with the other discussed methods.

本文研究了移动场景下圆形可重构智能表面（RIS）辅助毫米波（mmWave）多输入多输出非正交多址（MIMO-NOMA）系统的下行信道估计问题。首先，我们提出了一种子帧划分方案，以便将接收信号建模为满足正则多进分解（CPD）形式的四阶张量，从而将信道估计问题表述为张量分解和参数提取问题。然后，利用接收信号的全局和局部低秩特性，提出了一种双低秩四维张量分解模型，将接收信号分解为四个因子矩阵，并通过交替方向乘子法（ADMM）高效求解。随后，我们提出了一种基于Jacobi-Anger展开和圆形RIS特殊结构的两阶段参数估计方法来唯一解耦角度参数。此外，时延、多普勒频移和信道增益参数的估计也没有歧义，可以有效地提高其估计精度，特别是在低信噪比（SNR）下。最后，导出了cram - rao边界（CRB）的简洁封闭表达式作为性能基准。通过数值实验验证了该方法的有效性。

{"title":"Double low-rank 4D tensor decomposition for circular RIS-aided mmWave MIMO-NOMA system channel estimation in mobility scenarios","authors":"Wanyuan Cai , Youming Li , Menglei Sheng , Mingjun Huang , Qinke Qi , Shunli Hong","doi":"10.1016/j.sigpro.2025.110466","DOIUrl":"10.1016/j.sigpro.2025.110466","url":null,"abstract":"<div><div>In this paper, we consider a downlink channel estimation problem for circular reconfigurable intelligent surface (RIS)-aided millimeter-wave (mmWave) multiple-input multiple-output non-orthogonal multiple access (MIMO-NOMA) system in mobility scenarios. First, we propose a subframe partitioning scheme to facilitate the modeling of the received signal as a fourth-order tensor satisfying a canonical polyadic decomposition (CPD) form, thereby formulating the channel estimation problem as tensor decomposition and parameter extraction problems. Then, by exploiting both the global and local low-rank properties of the received signal, we propose a double low-rank 4D tensor decomposition model to decompose the received signal into four factor matrices, which is efficiently solved via alternating direction method of multipliers (ADMM). Subsequently, we propose a two-stage parameter estimation method based on the Jacobi-Anger expansion and the special structure of circular RIS to uniquely decouple the angle parameters. Furthermore, the time delay, Doppler shift, and channel gain parameters can also be estimated without ambiguities, and their estimation accuracy can be efficiently improved, especially at low signal-to-noise ratio (SNR). Finally, a concise closed-form expression for the Cramér-Rao bound (CRB) is derived as a performance benchmark. Numerical experiments are conducted to demonstrate the effectiveness of the proposed method compared with the other discussed methods.</div></div>","PeriodicalId":49523,"journal":{"name":"Signal Processing","volume":"243 ","pages":"Article 110466"},"PeriodicalIF":3.6,"publicationDate":"2025-12-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145885927","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

An efficient algorithm for PAPR minimization in OFDM-based joint radar-communication systems 基于ofdm的联合雷达通信系统中PAPR最小化的有效算法

IF 3.6 2区工程技术 Q2 ENGINEERING, ELECTRICAL & ELECTRONIC

Signal Processing

Pub Date : 2025-12-27 DOI: 10.1016/j.sigpro.2025.110474

Prasanth Logaraman, Aakash Arora, Prabhu Babu

In this paper, we propose an efficient algorithm to reduce the high peak-to-average power ratio (PAPR) in a quadrature amplitude modulation (QAM) orthogonal frequency division multiplexing (OFDM) based joint radar-communication (JRC) system. Due to the non-linear power amplifier at the OFDM transmitter, the waveform with high PAPR will be distorted, resulting in degraded sensing and communication performance. The formulated PAPR minimization problem is a fractional and non-convex optimization problem. To efficiently solve this problem, we use an iterative approach based on Dinkelbach’s method. However, the sub-problem at each iteration is non-convex. We use the majorization-minimization (MM) principle to handle this non-convex sub-problem. We then provide numerical simulations to compare the performance of the proposed algorithm with some of the recent methods in the literature. The proposed algorithm converges monotonically to a lower PAPR, achieves faster convergence in fewer iterations, and demonstrates good radar sensing and communication performance compared to recent methods.

在本文中，我们提出了一种有效的算法来降低正交调幅（QAM）正交频分复用（OFDM）联合雷达通信（JRC）系统的峰值平均功率比（PAPR）。由于OFDM发射机处的非线性功率放大器，高PAPR的波形会发生畸变，导致传感和通信性能下降。所建立的PAPR最小化问题是一个分数型非凸优化问题。为了有效地解决这一问题，我们采用了基于Dinkelbach方法的迭代方法。然而，每次迭代的子问题都是非凸的。我们使用最大化最小化（MM）原理来处理这个非凸子问题。然后，我们提供数值模拟来比较所提出的算法与文献中一些最新方法的性能。该算法单调收敛到较低的PAPR，迭代次数少，收敛速度快，与现有方法相比，具有良好的雷达感知和通信性能。

引用次数: 0

Robust beamforming for MIMO-RIS systems with hardware impairments 具有硬件缺陷的MIMO-RIS系统的鲁棒波束形成

IF 3.6 2区工程技术 Q2 ENGINEERING, ELECTRICAL & ELECTRONIC

Signal Processing

Pub Date : 2025-12-25 DOI: 10.1016/j.sigpro.2025.110460

Kunze Wu , Zhengyi Zhang , Jingya Ren, Chenglin Wang, Shiyong Chen, Weiheng Jiang

This paper presents a robust beamforming framework for multiple-input multiple-output (MIMO) systems enhanced by reconfigurable intelligent surfaces (RIS), accounting for practical hardware impairments. The system model incorporates key non-idealities, including low-resolution analog-to-digital converters (ADCs) and hybrid radio frequency (RF) chains affected by distortion. The central objective is to minimize the user-side transmit power through joint optimization of the analog and digital combiners, the RIS reffection coefffcients, and the transmit power level itself. Due to the high dimensionality and inherent nonconvexity of the formulated problem, we employ an alternating optimization (AO) scheme to partition the variables and simplify the solution process. Fractional programming (FP) is applied to derive closed-form expressions for the auxiliary variables, while the digital combiner is obtained using the Lagrangian multiplier technique. To address the optimization of the analog combiner and RIS conffguration, we further introduce the penalty dual decomposition (PDD) method. Simulation results confirm that the proposed design significantly outperforms baseline methods in reducing transmit power, even in the presence of hardware degradation. Moreover, the proposed algorithm exhibits rapid convergence and scalability across varying system conffgurations.

本文提出了一种多输入多输出（MIMO）系统的鲁棒波束形成框架，该框架通过可重构智能曲面（RIS）增强，考虑到实际硬件缺陷。该系统模型包含了关键的非理想特性，包括低分辨率模数转换器（adc）和受失真影响的混合射频（RF）链。中心目标是通过联合优化模拟和数字合成器、RIS反射系数和发射功率水平本身来最小化用户侧发射功率。由于该问题的高维性和固有的非凸性，我们采用交替优化（AO）格式对变量进行划分，简化了求解过程。采用分数规划方法推导辅助变量的封闭表达式，采用拉格朗日乘法器技术得到数字组合。为了解决模拟合成器和RIS配置的优化问题，我们进一步引入了惩罚对偶分解（PDD）方法。仿真结果证实，即使在存在硬件退化的情况下，所提出的设计在降低发射功率方面也明显优于基线方法。此外，该算法在不同的系统配置中具有快速收敛和可扩展性。

{"title":"Robust beamforming for MIMO-RIS systems with hardware impairments","authors":"Kunze Wu , Zhengyi Zhang , Jingya Ren, Chenglin Wang, Shiyong Chen, Weiheng Jiang","doi":"10.1016/j.sigpro.2025.110460","DOIUrl":"10.1016/j.sigpro.2025.110460","url":null,"abstract":"<div><div>This paper presents a robust beamforming framework for multiple-input multiple-output (MIMO) systems enhanced by reconfigurable intelligent surfaces (RIS), accounting for practical hardware impairments. The system model incorporates key non-idealities, including low-resolution analog-to-digital converters (ADCs) and hybrid radio frequency (RF) chains affected by distortion. The central objective is to minimize the user-side transmit power through joint optimization of the analog and digital combiners, the RIS reffection coefffcients, and the transmit power level itself. Due to the high dimensionality and inherent nonconvexity of the formulated problem, we employ an alternating optimization (AO) scheme to partition the variables and simplify the solution process. Fractional programming (FP) is applied to derive closed-form expressions for the auxiliary variables, while the digital combiner is obtained using the Lagrangian multiplier technique. To address the optimization of the analog combiner and RIS conffguration, we further introduce the penalty dual decomposition (PDD) method. Simulation results confirm that the proposed design significantly outperforms baseline methods in reducing transmit power, even in the presence of hardware degradation. Moreover, the proposed algorithm exhibits rapid convergence and scalability across varying system conffgurations.</div></div>","PeriodicalId":49523,"journal":{"name":"Signal Processing","volume":"243 ","pages":"Article 110460"},"PeriodicalIF":3.6,"publicationDate":"2025-12-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145885931","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

ESLIM: Extended sparse learning via iterative minimization ESLIM：通过迭代最小化扩展稀疏学习

IF 3.6 2区工程技术 Q2 ENGINEERING, ELECTRICAL & ELECTRONIC

Signal Processing

Pub Date : 2025-12-25 DOI: 10.1016/j.sigpro.2025.110465

Avi Leikind, Ofer Amrani

Different algorithms were proposed in recent years for achieving the overall goal of identifying a sparse approximation of a measured signal with high accuracy and relatively low computational complexity. Of paramount interest to the current work is the known Sparse Learning via Iterative Minimization (SLIM) algorithm, which is a maximum a-posteriori (MAP)-based approach for sparse signal recovery, originally proposed for multiple-input multiple-output (MIMO) radar imaging. The current work aims at introducing a modification to SLIM, termed Extended Sparse Learning via Iterative Minimization (ESLIM). This includes formulation of an extended cost function and derivation of an iterative optimization backed by proof-of-convergence. The extended algorithm aims to provide accurate sparse signal estimates for various applications and in different settings, e.g., a correlated dictionary (that is, a collection of signals composed of elementary signals similar to “choose” from) and short observation times. The solutions provided by the proposed algorithm, demonstrating its superiority and accuracy, are compared to several state-of-the-art sparse recovery algorithms, while focusing on MIMO radar imaging.

近年来提出了不同的算法，以实现以高精度和相对低的计算复杂度识别测量信号的稀疏逼近的总体目标。当前工作最感兴趣的是已知的稀疏学习迭代最小化（SLIM）算法，这是一种基于最大后验（MAP）的稀疏信号恢复方法，最初提出用于多输入多输出（MIMO）雷达成像。目前的工作旨在引入SLIM的修改，称为通过迭代最小化扩展稀疏学习（ESLIM）。这包括一个扩展的成本函数的公式和迭代优化的推导支持收敛证明。扩展算法旨在为各种应用和不同设置提供准确的稀疏信号估计，例如，相关字典（即由类似于“选择”的基本信号组成的信号集合）和短观测时间。该算法提供的解决方案显示了其优越性和准确性，并与几种最先进的稀疏恢复算法进行了比较，同时重点关注MIMO雷达成像。

{"title":"ESLIM: Extended sparse learning via iterative minimization","authors":"Avi Leikind, Ofer Amrani","doi":"10.1016/j.sigpro.2025.110465","DOIUrl":"10.1016/j.sigpro.2025.110465","url":null,"abstract":"<div><div>Different algorithms were proposed in recent years for achieving the overall goal of identifying a sparse approximation of a measured signal with high accuracy and relatively low computational complexity. Of paramount interest to the current work is the known Sparse Learning via Iterative Minimization (SLIM) algorithm, which is a maximum a-posteriori (MAP)-based approach for sparse signal recovery, originally proposed for multiple-input multiple-output (MIMO) radar imaging. The current work aims at introducing a modification to SLIM, termed Extended Sparse Learning via Iterative Minimization (ESLIM). This includes formulation of an extended cost function and derivation of an iterative optimization backed by proof-of-convergence. The extended algorithm aims to provide accurate sparse signal estimates for various applications and in different settings, e.g., a correlated dictionary (that is, a collection of signals composed of elementary signals similar to “choose” from) and short observation times. The solutions provided by the proposed algorithm, demonstrating its superiority and accuracy, are compared to several state-of-the-art sparse recovery algorithms, while focusing on MIMO radar imaging.</div></div>","PeriodicalId":49523,"journal":{"name":"Signal Processing","volume":"243 ","pages":"Article 110465"},"PeriodicalIF":3.6,"publicationDate":"2025-12-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145885932","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Sense -assisted hybrid beamforming based on deep unfolding network for THz UM-MIMO-ISAC system 基于深度展开网络的太赫兹UM-MIMO-ISAC系统传感辅助混合波束形成

IF 3.6 2区工程技术 Q2 ENGINEERING, ELECTRICAL & ELECTRONIC

Signal Processing

Pub Date : 2025-12-24 DOI: 10.1016/j.sigpro.2025.110458

Yang Liu , Qiyue Chang , Yuan Xing , Shuhao Zhang , Qintuya Si , Tianshuang Qiu

Terahertz ultra-massive multiple-input multiple-output integrated sensing and communication (THz UM-MIMO-ISAC) is regarded as a promising enabler for future 6G wireless systems. However, its performance is significantly constrained by severe beam-splitting effects in near-field environments, which undermine beamforming accuracy and decrease spectral efficiency (SE). To address these challenges, this paper proposes a sensing-assisted hybrid beamforming scheme based on a deep unfolding network (DUN) for THz UM-MIMO-ISAC systems. A mixed-field dual-layer true-time-delay (TTD) architecture is designed to jointly model spherical and planar wavefronts, thereby alleviating near-field beam squint and array gain degradation while maintaining low complexity. To further improve beam alignment, echo-sensing signals are initially processed by the rooted multiple signal classification (Root-MUSIC) algorithm for initial angle of arrival (AoA) estimation, followed by a long short-term memory (LSTM) network that refines and predicts the AoA for channel reconstruction. The predicted AoA is subsequently embedded into a DUN-based hybrid beamforming framework, where iterative optimization is unfolded into a trainable structure that jointly designs analog and digital beamforming matrices using unsupervised learning. By leveraging Root-MUSIC for initial estimation, LSTM for temporal prediction, and DUN for adaptive optimization, the proposed method achieves synergy optimization. Simulation results show that the proposed framework achieves higher SE, improved beamforming accuracy and stability, and reduced computational complexity compared with state-of-the-art approaches.

太赫兹超大规模多输入多输出集成传感和通信（THz UM-MIMO-ISAC）被认为是未来6G无线系统的一个有前途的推动者。然而，在近场环境中，严重的波束分裂效应会影响波束形成的精度，降低波束形成的频谱效率（SE）。为了解决这些挑战，本文提出了一种基于深度展开网络（DUN）的传感辅助混合波束形成方案，用于太赫兹UM-MIMO-ISAC系统。设计了一种混合场双层真时延（TTD）架构，联合模拟球面波前和平面波前，从而在保持低复杂度的同时减轻近场波束斜视和阵列增益退化。为了进一步改善波束对准，回波传感信号首先通过根多信号分类（Root-MUSIC）算法进行初始到达角（AoA）估计，然后使用长短期记忆（LSTM）网络对AoA进行细化和预测，用于信道重建。预测的AoA随后嵌入到基于dun的混合波束形成框架中，其中迭代优化展开为可训练的结构，该结构使用无监督学习联合设计模拟和数字波束形成矩阵。该方法利用Root-MUSIC进行初始估计，LSTM进行时间预测，DUN进行自适应优化，实现了协同优化。仿真结果表明，与现有方法相比，该框架实现了更高的SE，提高了波束形成精度和稳定性，降低了计算复杂度。

{"title":"Sense -assisted hybrid beamforming based on deep unfolding network for THz UM-MIMO-ISAC system","authors":"Yang Liu , Qiyue Chang , Yuan Xing , Shuhao Zhang , Qintuya Si , Tianshuang Qiu","doi":"10.1016/j.sigpro.2025.110458","DOIUrl":"10.1016/j.sigpro.2025.110458","url":null,"abstract":"<div><div>Terahertz ultra-massive multiple-input multiple-output integrated sensing and communication (THz UM-MIMO-ISAC) is regarded as a promising enabler for future 6G wireless systems. However, its performance is significantly constrained by severe beam-splitting effects in near-field environments, which undermine beamforming accuracy and decrease spectral efficiency (SE). To address these challenges, this paper proposes a sensing-assisted hybrid beamforming scheme based on a deep unfolding network (DUN) for THz UM-MIMO-ISAC systems. A mixed-field dual-layer true-time-delay (TTD) architecture is designed to jointly model spherical and planar wavefronts, thereby alleviating near-field beam squint and array gain degradation while maintaining low complexity. To further improve beam alignment, echo-sensing signals are initially processed by the rooted multiple signal classification (Root-MUSIC) algorithm for initial angle of arrival (AoA) estimation, followed by a long short-term memory (LSTM) network that refines and predicts the AoA for channel reconstruction. The predicted AoA is subsequently embedded into a DUN-based hybrid beamforming framework, where iterative optimization is unfolded into a trainable structure that jointly designs analog and digital beamforming matrices using unsupervised learning. By leveraging Root-MUSIC for initial estimation, LSTM for temporal prediction, and DUN for adaptive optimization, the proposed method achieves synergy optimization. Simulation results show that the proposed framework achieves higher SE, improved beamforming accuracy and stability, and reduced computational complexity compared with state-of-the-art approaches.</div></div>","PeriodicalId":49523,"journal":{"name":"Signal Processing","volume":"243 ","pages":"Article 110458"},"PeriodicalIF":3.6,"publicationDate":"2025-12-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145885933","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Weak and small target detection algorithm based on parallel infrared polarization feature estimation 基于并行红外偏振特征估计的弱小目标检测算法

IF 3.6 2区工程技术 Q2 ENGINEERING, ELECTRICAL & ELECTRONIC

Signal Processing

Pub Date : 2025-12-24 DOI: 10.1016/j.sigpro.2025.110464

Cailing Zhao , Zhiguo Fan , Yunyou Hu , Hongli Jiang , Yunxiang Zhang

Infrared polarization is an image technique that combines the advantage of both polarization imaging and infrared imaging. With the help of polarization latitude information, infrared polarization can significantly improve the ability of target detection and identification by enhancing the differences between the target and the background. Nevertheless, polarization imaging can be easily interfered by cloud edges and bright clutter in the sky cloud background. For this problem, we propose a weak and small target detection algorithm based on parallel infrared polarization feature estimation. The parallel means different processing approaches for the polarimetric images with different polarization features. Firstly, the cloud background edge clutter may appear local autocorrelation with strong pixel continuity in the two polarimetric images Q and U. On this account, a 4-Direction Anisotropic Convolutional Filter Bank (4D-ACFB) is proposed to remove the cloud background through the directional features in the images captured by filters in various locations. Secondly, for the weak saliency of the target in the Degree of Linear Polarization (DoLP) cloud background, a Point Spread Function (PSF) is introduced to correct the 4-direction anisotropic convolutional filter kernel to enhance the target. Finally, for the filtered images mentioned above, the Sparse Regularized Optimization (SRO) is proposed to remove residual clutter. In this paper, the effectiveness of the proposed algorithm has been demonstrated by experimental data analysis of the actual collected images. Comparing with other algorithms, this algorithm can effectively enhance the target edge information and improve the performance of robust detection, while suppressing the cloud background edges and highlighted clutter at the same time.

红外偏振成像是一种结合了偏振成像和红外成像优点的成像技术。借助偏振纬度信息，红外偏振通过增强目标与背景的差异，可以显著提高目标的检测和识别能力。然而，极化成像容易受到云边缘和天云背景中明亮杂波的干扰。针对这一问题，提出了一种基于并行红外偏振特征估计的弱小目标检测算法。并行意味着对具有不同偏振特征的偏振图像采用不同的处理方法。首先，在两幅偏振图像Q和u中，云背景边缘杂波可能出现局部自相关且像素连续性强的问题，为此，提出了一种4-Direction各向异性卷积滤波器组（4D-ACFB），通过滤波器在不同位置捕获的图像中的方向特征去除云背景。其次，针对目标在线性极化度（DoLP）云背景下的弱显著性，引入点扩散函数（PSF）对四向各向异性卷积滤波核进行校正，增强目标；最后，针对上述滤波后的图像，提出了稀疏正则化优化方法（SRO）去除残留杂波。本文通过对实际采集图像的实验数据分析，验证了该算法的有效性。与其他算法相比，该算法能够有效增强目标边缘信息，提高鲁棒检测性能，同时抑制云背景边缘和突出杂波。

{"title":"Weak and small target detection algorithm based on parallel infrared polarization feature estimation","authors":"Cailing Zhao , Zhiguo Fan , Yunyou Hu , Hongli Jiang , Yunxiang Zhang","doi":"10.1016/j.sigpro.2025.110464","DOIUrl":"10.1016/j.sigpro.2025.110464","url":null,"abstract":"<div><div>Infrared polarization is an image technique that combines the advantage of both polarization imaging and infrared imaging. With the help of polarization latitude information, infrared polarization can significantly improve the ability of target detection and identification by enhancing the differences between the target and the background. Nevertheless, polarization imaging can be easily interfered by cloud edges and bright clutter in the sky cloud background. For this problem, we propose a weak and small target detection algorithm based on parallel infrared polarization feature estimation. The parallel means different processing approaches for the polarimetric images with different polarization features. Firstly, the cloud background edge clutter may appear local autocorrelation with strong pixel continuity in the two polarimetric images Q and U. On this account, a 4-Direction Anisotropic Convolutional Filter Bank (4D-ACFB) is proposed to remove the cloud background through the directional features in the images captured by filters in various locations. Secondly, for the weak saliency of the target in the Degree of Linear Polarization (DoLP) cloud background, a Point Spread Function (PSF) is introduced to correct the 4-direction anisotropic convolutional filter kernel to enhance the target. Finally, for the filtered images mentioned above, the Sparse Regularized Optimization (SRO) is proposed to remove residual clutter. In this paper, the effectiveness of the proposed algorithm has been demonstrated by experimental data analysis of the actual collected images. Comparing with other algorithms, this algorithm can effectively enhance the target edge information and improve the performance of robust detection, while suppressing the cloud background edges and highlighted clutter at the same time.</div></div>","PeriodicalId":49523,"journal":{"name":"Signal Processing","volume":"243 ","pages":"Article 110464"},"PeriodicalIF":3.6,"publicationDate":"2025-12-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145885924","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

An information geometry interpretation for approximate message passing 近似消息传递的信息几何解释

IF 3.6 2区工程技术 Q2 ENGINEERING, ELECTRICAL & ELECTRONIC

Signal Processing

Pub Date : 2025-12-23 DOI: 10.1016/j.sigpro.2025.110462

Bingyan Liu , An-An Lu , Mingrui Fan , Jiyuan Yang , Xiqi Gao

In this paper, a novel information geometry (IG) framework to solve the standard linear regression problem with non-Gaussian a priori distribution is proposed. The proposed framework is also simpler than that in previous works when the a priori distribution becomes Gaussian. By applying the framework, a new information geometry approach (IGA) for the basis pursuit de-noising (BPDN) in standard linear regression is derived. Its convergence behavior is then analyzed. To establish the relation between the IGA and the approximate message passing (AMP) algorithm, the approximate information geometry approach (AIGA) for BPDN is derived from the IGA, and proved to be equivalent to the AMP algorithm. We also show how the algorithm derived from the IG framework relates to the generalized AMP (GAMP) and vector AMP (VAMP). These intrinsic results offer a new perspective for the AMP algorithm, and clues for understanding and improving stochastic reasoning methods.

本文提出了一种新的信息几何框架来解决非高斯先验分布的标准线性回归问题。当先验分布变为高斯分布时，所提出的框架也比以前的工作更简单。应用该框架，推导了一种新的用于标准线性回归中基追踪去噪的信息几何方法（IGA）。然后分析了其收敛性。为了建立IGA与近似消息传递（AMP）算法之间的关系，在IGA的基础上推导出BPDN的近似信息几何方法（AIGA），并证明其等价于AMP算法。我们还展示了从IG框架导出的算法如何与广义AMP （GAMP）和向量AMP （VAMP）相关。这些内在结果为AMP算法提供了新的视角，也为理解和改进随机推理方法提供了线索。

引用次数: 0

Activity-dependent resolution adjustment for radar-based human activity recognition 基于雷达的人类活动识别的活动相关分辨率调整

IF 3.6 2区工程技术 Q2 ENGINEERING, ELECTRICAL & ELECTRONIC

Signal Processing

Pub Date : 2025-12-21 DOI: 10.1016/j.sigpro.2025.110456

Do-Hyun Park, Min-Wook Jeon, Hyoung-Nam Kim

The rising demand for detecting hazardous situations has led to increased interest in radar-based human activity recognition (HAR). Conventional radar-based HAR methods predominantly rely on micro-Doppler spectrograms for recognition tasks. However, conventional spectrograms employ a fixed resolution regardless of the varying characteristics of human activities, leading to limited representation of micro-Doppler signatures. To address this limitation, we propose a time-frequency domain representation method that adaptively adjusts the resolution based on activity characteristics. This approach adaptively adjusts the spectrogram resolution in a nonlinear manner, emphasizing frequency ranges that vary with activity intensity and are critical to capturing micro-Doppler signatures. We validate the proposed method by training deep learning-based HAR models on datasets generated using our adaptive representation. Experimental results demonstrate that models trained with our method achieve superior recognition accuracy compared to those trained with conventional methods.

探测危险情况的需求不断增长，导致人们对基于雷达的人类活动识别（HAR）的兴趣增加。传统的基于雷达的HAR方法主要依靠微多普勒谱图进行识别任务。然而，传统的频谱图采用固定的分辨率，而不考虑人类活动的不同特征，导致微多普勒特征的有限表示。为了解决这一限制，我们提出了一种基于活动特征自适应调整分辨率的时频域表示方法。该方法以非线性方式自适应调整频谱图分辨率，强调随活动强度变化的频率范围，对于捕获微多普勒特征至关重要。我们通过在使用自适应表示生成的数据集上训练基于深度学习的HAR模型来验证所提出的方法。实验结果表明，与传统方法训练的模型相比，用该方法训练的模型具有更高的识别精度。

引用次数: 0