Signal Processing最新文献

英文中文

Incremental-based FxLMS algorithm for distributed multichannel narrowband ANC system and its performance analysis 分布式多信道窄带ANC系统中基于增量的FxLMS算法及其性能分析

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

Signal Processing

Pub Date : 2025-12-14 DOI: 10.1016/j.sigpro.2025.110436

Jing Chen, Xiaolei Li, Wei Gao

Sinusoidal noise generated by rotating machinery can be effectively attenuated using narrowband active noise control (NANC) systems based on the filtered-x least mean square (FxLMS) algorithm. While multichannel NANC (MNANC) systems achieve satisfactory performance in large spatial areas, they impose a heavy computational burden. To address this issue, this paper proposes an incremental-based FxLMS (INFxLMS) algorithm for distributed NANC systems, which efficiently distributes the computational load across network nodes. A partial-update INFxLMS algorithm (p-INFxLMS) is also introduced to accommodate communication latency in practical distributed networks. Moreover, a detailed performance analysis-covering both transient and steady-state behavior is provided to characterize the convergence properties of the proposed algorithms. Simulation results demonstrate that in a latency-free network, the INFxLMS algorithm delivers noise reduction performance comparable to that of the centralized approach. And the p-INFxLMS algorithm which is more aligned with reality can achieve good noise reduction while greatly reducing the computational burden. Additionally, computer simulations validate the accuracy of the theoretical analysis.

基于滤波最小均方（FxLMS）算法的窄带有源噪声控制（NANC）系统可以有效地衰减旋转机械产生的正弦噪声。虽然多通道纳米控制（MNANC）系统在大空间范围内取得了令人满意的性能，但它们带来了沉重的计算负担。为了解决这一问题，本文提出了一种基于增量的FxLMS （INFxLMS）算法，该算法有效地将计算负荷分配到网络节点上。为了适应实际分布式网络中的通信延迟，还引入了部分更新的INFxLMS算法（p-INFxLMS）。此外，还提供了详细的性能分析，包括瞬态和稳态行为，以表征所提出算法的收敛特性。仿真结果表明，在无延迟网络中，INFxLMS算法的降噪性能与集中式方法相当。而p-INFxLMS算法更符合实际，在大大减少计算量的同时，也能达到较好的降噪效果。此外，计算机模拟验证了理论分析的准确性。

{"title":"Incremental-based FxLMS algorithm for distributed multichannel narrowband ANC system and its performance analysis","authors":"Jing Chen, Xiaolei Li, Wei Gao","doi":"10.1016/j.sigpro.2025.110436","DOIUrl":"10.1016/j.sigpro.2025.110436","url":null,"abstract":"<div><div>Sinusoidal noise generated by rotating machinery can be effectively attenuated using narrowband active noise control (NANC) systems based on the filtered-x least mean square (FxLMS) algorithm. While multichannel NANC (MNANC) systems achieve satisfactory performance in large spatial areas, they impose a heavy computational burden. To address this issue, this paper proposes an incremental-based FxLMS (INFxLMS) algorithm for distributed NANC systems, which efficiently distributes the computational load across network nodes. A partial-update INFxLMS algorithm (p-INFxLMS) is also introduced to accommodate communication latency in practical distributed networks. Moreover, a detailed performance analysis-covering both transient and steady-state behavior is provided to characterize the convergence properties of the proposed algorithms. Simulation results demonstrate that in a latency-free network, the INFxLMS algorithm delivers noise reduction performance comparable to that of the centralized approach. And the p-INFxLMS algorithm which is more aligned with reality can achieve good noise reduction while greatly reducing the computational burden. Additionally, computer simulations validate the accuracy of the theoretical analysis.</div></div>","PeriodicalId":49523,"journal":{"name":"Signal Processing","volume":"242 ","pages":"Article 110436"},"PeriodicalIF":3.6,"publicationDate":"2025-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145790806","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

Infrared and visible image fusion via spatial-frequency edge-aware network 基于空频边缘感知网络的红外与可见光图像融合

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

Signal Processing

Pub Date : 2025-12-13 DOI: 10.1016/j.sigpro.2025.110441

Shuohui Li , Qilei Li , Mingliang Gao , Lucia Cascone , Dan Zhang

The objective of combining infrared with visible images lies in merging essential visual data from both sources to produce an enhanced output. Existing fusion methods predominantly operate within the spatial domain, while ignoring valuable data that could be extracted from the frequency domain. Therefore, the fusion performance remains suboptimal. To overcome this drawback, we introduce the Spatial-Frequency Edge-Aware Network(SFEANet) model, which employs a parallel dual-branch structure that simultaneously processes spatial and frequency domain information. The spatial fusion branch utilizes the Edge Feature Extraction(EFE) block and the Self Attention(SA) block to capture and integrate key features across both image types. The frequency-domain fusion branch first applies the Fast Fourier Transform(FFT) for domain conversion, which transforms the input into spectral representations. Subsequently, it performs interactive operations on their amplitude and phase components to enable cross-modal feature integration. The fused features are ultimately reconstructed in the spatial domain through the Inverse Fast Fourier Transform (IFFT). Comprehensive experiments conducted on three public benchmarks demonstrate the superior performance of SFEANet across multiple quantitative measures and perceptual quality assessments. The implementation can be accessed via https://github.com/lishuohui123/SFEANet.

将红外图像与可见光图像相结合的目的在于合并来自两个来源的基本视觉数据，以产生增强的输出。现有的融合方法主要在空间域内操作，而忽略了可以从频域提取的有价值的数据。因此，融合性能仍然是次优的。为了克服这一缺点，我们引入了空间-频率边缘感知网络（SFEANet）模型，该模型采用并行双分支结构同时处理空间和频域信息。空间融合分支利用边缘特征提取（EFE）块和自关注（SA）块来捕获和整合两种图像类型的关键特征。频域融合分支首先应用快速傅里叶变换（FFT）进行域转换，将输入转换为频谱表示。随后，它对它们的幅度和相位分量进行交互操作，以实现跨模态特征集成。最后通过快速傅里叶反变换（IFFT）在空间域中重构融合特征。在三个公共基准上进行的综合实验表明，SFEANet在多个定量测量和感知质量评估方面表现优异。实现可以通过https://github.com/lishuohui123/SFEANet访问。

{"title":"Infrared and visible image fusion via spatial-frequency edge-aware network","authors":"Shuohui Li , Qilei Li , Mingliang Gao , Lucia Cascone , Dan Zhang","doi":"10.1016/j.sigpro.2025.110441","DOIUrl":"10.1016/j.sigpro.2025.110441","url":null,"abstract":"<div><div>The objective of combining infrared with visible images lies in merging essential visual data from both sources to produce an enhanced output. Existing fusion methods predominantly operate within the spatial domain, while ignoring valuable data that could be extracted from the frequency domain. Therefore, the fusion performance remains suboptimal. To overcome this drawback, we introduce the Spatial-Frequency Edge-Aware Network(SFEANet) model, which employs a parallel dual-branch structure that simultaneously processes spatial and frequency domain information. The spatial fusion branch utilizes the Edge Feature Extraction(EFE) block and the Self Attention(SA) block to capture and integrate key features across both image types. The frequency-domain fusion branch first applies the Fast Fourier Transform(FFT) for domain conversion, which transforms the input into spectral representations. Subsequently, it performs interactive operations on their amplitude and phase components to enable cross-modal feature integration. The fused features are ultimately reconstructed in the spatial domain through the Inverse Fast Fourier Transform (IFFT). Comprehensive experiments conducted on three public benchmarks demonstrate the superior performance of SFEANet across multiple quantitative measures and perceptual quality assessments. The implementation can be accessed via <span><span>https://github.com/lishuohui123/SFEANet</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":49523,"journal":{"name":"Signal Processing","volume":"242 ","pages":"Article 110441"},"PeriodicalIF":3.6,"publicationDate":"2025-12-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145790811","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

Multi-Shaft Speed-Informed Adaptive Window Filtering Method for Acoustic Pressure Signals in Marine Gas Turbines 船用燃气轮机声压信号的多轴转速自适应窗滤波方法

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

Signal Processing

Pub Date : 2025-12-12 DOI: 10.1016/j.sigpro.2025.110448

Xiaoyu Han , Yunpeng Cao , Minghao Wu , Weiying Wang , Weixing Feng

As the core component of marine propulsion systems, the operational condition of gas turbines plays a critical role in ensuring the overall reliability of the vessel. Compared with traditional vibration signals, acoustic pressure signals offer advantages such as non-contact measurement, easy installation, and wide spatial coverage, exhibiting higher sensing sensitivity in complex installation environments. However, under non-stationary operating conditions, acoustic pressure signals are highly susceptible to high-frequency noise and transient disturbances, which significantly compromise the accuracy of equipment condition monitoring. To address this issue, this paper proposes a multi-shaft speed-informed adaptive window Savitzky-Golay filtering algorithm (MSSI-AWSG). Operating within a distributed sensing framework, the method utilizes multi-point acquisition of shaft rotational speeds and acoustic pressure signals, we construct a cross-modal physical perception model. A composite window-scaling mechanism is further introduced to enable real-time adaptive adjustment of the filtering window. In addition, a residual enhancement module is designed, integrating robust statistical techniques and proportional compression strategies to effectively preserve transient high-frequency disturbance features. The algorithm is deployed on edge computing nodes to perform real-time filtering at the data acquisition side, thereby reducing downstream storage and computational burdens. Experiments based on real-world distributed shaft speed and acoustic pressure data demonstrate that the proposed method outperforms existing approaches in terms of acoustic signal fidelity, transient response, and noise suppression. The results verify the feasibility of this approach for building shipborne distributed online acoustic signal processing systems and their engineering applicability.

燃气轮机作为船舶推进系统的核心部件，其运行状态对保证船舶整体可靠性起着至关重要的作用。与传统的振动信号相比，声压信号具有非接触式测量、安装方便、空间覆盖范围广等优点，在复杂的安装环境中具有更高的传感灵敏度。然而，在非平稳工况下，声压信号极易受到高频噪声和瞬态干扰的影响，严重影响设备状态监测的准确性。为了解决这一问题，本文提出了一种多轴速度通知自适应窗口Savitzky-Golay滤波算法（MSSI-AWSG）。该方法在分布式传感框架内运行，利用多点采集轴转速和声压信号，构建了一个跨模态物理感知模型。进一步引入了复合窗口缩放机制，实现了滤波窗口的实时自适应调整。此外，设计了残差增强模块，集成了鲁棒统计技术和比例压缩策略，有效地保留了瞬态高频干扰特征。该算法部署在边缘计算节点上，在数据采集端进行实时过滤，减少下游存储和计算负担。基于真实分布轴速和声压数据的实验表明，该方法在声信号保真度、瞬态响应和噪声抑制方面优于现有方法。仿真结果验证了该方法在构建舰载分布式在线声信号处理系统中的可行性和工程适用性。

{"title":"Multi-Shaft Speed-Informed Adaptive Window Filtering Method for Acoustic Pressure Signals in Marine Gas Turbines","authors":"Xiaoyu Han , Yunpeng Cao , Minghao Wu , Weiying Wang , Weixing Feng","doi":"10.1016/j.sigpro.2025.110448","DOIUrl":"10.1016/j.sigpro.2025.110448","url":null,"abstract":"<div><div>As the core component of marine propulsion systems, the operational condition of gas turbines plays a critical role in ensuring the overall reliability of the vessel. Compared with traditional vibration signals, acoustic pressure signals offer advantages such as non-contact measurement, easy installation, and wide spatial coverage, exhibiting higher sensing sensitivity in complex installation environments. However, under non-stationary operating conditions, acoustic pressure signals are highly susceptible to high-frequency noise and transient disturbances, which significantly compromise the accuracy of equipment condition monitoring. To address this issue, this paper proposes a multi-shaft speed-informed adaptive window Savitzky-Golay filtering algorithm (MSSI-AWSG). Operating within a distributed sensing framework, the method utilizes multi-point acquisition of shaft rotational speeds and acoustic pressure signals, we construct a cross-modal physical perception model. A composite window-scaling mechanism is further introduced to enable real-time adaptive adjustment of the filtering window. In addition, a residual enhancement module is designed, integrating robust statistical techniques and proportional compression strategies to effectively preserve transient high-frequency disturbance features. The algorithm is deployed on edge computing nodes to perform real-time filtering at the data acquisition side, thereby reducing downstream storage and computational burdens. Experiments based on real-world distributed shaft speed and acoustic pressure data demonstrate that the proposed method outperforms existing approaches in terms of acoustic signal fidelity, transient response, and noise suppression. The results verify the feasibility of this approach for building shipborne distributed online acoustic signal processing systems and their engineering applicability.</div></div>","PeriodicalId":49523,"journal":{"name":"Signal Processing","volume":"243 ","pages":"Article 110448"},"PeriodicalIF":3.6,"publicationDate":"2025-12-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145802319","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

Robust DOA estimation with combined adaptive filtering and alternating maximum versoria based denoising 结合自适应滤波和交替最大模量去噪的鲁棒DOA估计

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

Signal Processing

Pub Date : 2025-12-11 DOI: 10.1016/j.sigpro.2025.110439

Yijie Tang , Jun Peng , Ying-Ren Chien , Hao Zeng

In this communication, we aim to address the challenge that existing adaptive filtering-based direction of arrival (DOA) estimation methods struggle to perform reliably in harsh environments (e.g., low signal-to-noise ratio (SNR), impulsive noise, limited array elements or snapshots, etc.). To overcome this, we propose a robust alternating maximum versoria based denoising (AMVD) approach for preprocessing to partially mitigate noise. The proposed AMVD approach inherently suppresses outlier-induced distortions, ensuring robustness in both Gaussian and impulsive noise scenarios. Subsequently, we design a combined adaptive filtering scheme comprising two independent filters to construct an orthogonal space for the array manifold vectors corresponding to the source signals, enabling the derivation of a high-performance spatial spectrum. Comprehensive simulations demonstrate that the proposed algorithm outperforms state-of-the-art techniques.

在本通信中，我们的目标是解决现有的基于自适应滤波的到达方向（DOA）估计方法在恶劣环境（例如，低信噪比（SNR），脉冲噪声，有限阵列元素或快照等）中难以可靠执行的挑战。为了克服这个问题，我们提出了一种鲁棒的基于交替最大模量的去噪（AMVD）方法进行预处理，以部分减轻噪声。提出的AMVD方法固有地抑制了异常值引起的失真，确保了在高斯和脉冲噪声情况下的鲁棒性。随后，我们设计了一个由两个独立滤波器组成的组合自适应滤波方案，为源信号对应的阵列流形向量构建一个正交空间，从而能够推导出高性能的空间频谱。综合仿真表明，所提出的算法优于最先进的技术。

引用次数: 0

VQ-CSA: Addressing overgeneralization in video anomaly detection via contrastive feature discretization VQ-CSA：利用对比特征离散化解决视频异常检测中的过度泛化问题

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

Signal Processing

Pub Date : 2025-12-11 DOI: 10.1016/j.sigpro.2025.110442

Bolin Xiao, Wenjun Zhou, Rui Qu, Jiachen Dang, Quan Zhang, Bo Peng

Video anomaly detection (VAD) plays a critical role in public safety, industrial inspection, and traffic monitoring. Most unsupervised approaches rely on one-class classification to model normal patterns; however, existing generative models often suffer from overgeneralization, which can degrade anomaly detection performance. To address this limitation, we propose a reconstruction-based framework, VQ-CSA, which integrates Contrastive Skip-Attention (CSA) with Hierarchical Vector Quantization (VQ) to enhance anomaly discrimination while mitigating over-reconstruction. Specifically, CSA enhances feature selectivity and suppresses extraneous information, while VQ discretizes semantic representations to improve the discrimination of anomalies. Additionally, a multi-perspective anomaly scoring strategy is introduced, which combines reconstruction, structural gradient, and quantization errors to facilitate robust frame-level anomaly detection. Experimental results on the UCSD Ped2, CUHK Avenue, and ShanghaiTech datasets demonstrate that VQ-CSA achieves state-of-the-art performance, with AUCs of 98.6 %, 90.3 %, and 74.7 %, respectively, under a fully unsupervised and lightweight design.

视频异常检测（VAD）在公共安全、工业检查、交通监控等领域发挥着重要作用。大多数无监督方法依赖于单类分类来模拟正常模式；然而，现有的生成模型往往存在过度泛化的问题，这会降低异常检测的性能。为了解决这一问题，我们提出了一个基于重建的框架VQ-CSA，该框架将对比跳过注意（CSA）和层次向量量化（VQ）相结合，以增强异常识别，同时减少过度重建。具体而言，CSA增强了特征选择性并抑制了无关信息，而VQ离散化语义表示以提高异常的识别能力。此外，提出了一种结合重构、结构梯度和量化误差的多视角异常评分策略，以实现鲁棒的帧级异常检测。在UCSD Ped2、CUHK Avenue和ShanghaiTech数据集上的实验结果表明，在完全无监督和轻量化设计下，VQ-CSA达到了最先进的性能，auc分别为98.6%、90.3%和74.7%。

{"title":"VQ-CSA: Addressing overgeneralization in video anomaly detection via contrastive feature discretization","authors":"Bolin Xiao, Wenjun Zhou, Rui Qu, Jiachen Dang, Quan Zhang, Bo Peng","doi":"10.1016/j.sigpro.2025.110442","DOIUrl":"10.1016/j.sigpro.2025.110442","url":null,"abstract":"<div><div>Video anomaly detection (VAD) plays a critical role in public safety, industrial inspection, and traffic monitoring. Most unsupervised approaches rely on one-class classification to model normal patterns; however, existing generative models often suffer from overgeneralization, which can degrade anomaly detection performance. To address this limitation, we propose a reconstruction-based framework, VQ-CSA, which integrates Contrastive Skip-Attention (CSA) with Hierarchical Vector Quantization (VQ) to enhance anomaly discrimination while mitigating over-reconstruction. Specifically, CSA enhances feature selectivity and suppresses extraneous information, while VQ discretizes semantic representations to improve the discrimination of anomalies. Additionally, a multi-perspective anomaly scoring strategy is introduced, which combines reconstruction, structural gradient, and quantization errors to facilitate robust frame-level anomaly detection. Experimental results on the UCSD Ped2, CUHK Avenue, and ShanghaiTech datasets demonstrate that VQ-CSA achieves state-of-the-art performance, with AUCs of 98.6 %, 90.3 %, and 74.7 %, respectively, under a fully unsupervised and lightweight design.</div></div>","PeriodicalId":49523,"journal":{"name":"Signal Processing","volume":"242 ","pages":"Article 110442"},"PeriodicalIF":3.6,"publicationDate":"2025-12-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145790810","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

Conditional diffusion model for infrared and visible image fusion in open environments with few denoising steps 开放环境下红外与可见光图像融合的条件扩散模型，去噪步骤少

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

Signal Processing

Pub Date : 2025-12-11 DOI: 10.1016/j.sigpro.2025.110437

Luojie Yang , Chunming Li , Meng Yu , Guangyan Chen , Yufeng Yue

Multi-modality fusion improves perceptual robustness and accuracy by leveraging multi-source sensor data. Current RGB-T fusion methods still falter with adverse illumination and weather. Recent advances in generative methods especially diffusion models have shown the ability to enhance visible images under adverse conditions. However, the fusion of RGB-T still suffer from cross-modal feature loss, sensitivity to environmental interference, and prolonged generation times. These limitations arise due to: (1) difficulties in sufficiently extracting modality-specific information only within shared forward networks; (2) neglecting the interference from adverse weather conditions; (3) the multi-step denoising process in diffusion-based models, which increases temporal cost. To overcome these challenges, we propose a novel conditional diffusion model for RGB-T image fusion, named CDMFusion, which incorporates: (1) a three-branch network designed for fusion to more fully preserve information; (2) a multi-scene adaptive feature enhancer that dynamically enhances valuable features while mitigating interference; (3) a novel skip patrol mechanism enabling high-quality generation via two-step denoising without extra training. Additionally, a new multi-scene RGB-T image dataset and a dataset with multi-interference are released for comprehensive evaluation. Experiments demonstrate our method achieves superior performance across 7 datasets compared to 14 state-of-the-art methods. Code and datasets are at https://github.com/yangluojie/CDM.

多模态融合通过利用多源传感器数据提高感知鲁棒性和准确性。目前的RGB-T融合方法仍然受到恶劣光照和天气的影响。生成方法的最新进展，特别是扩散模型已经显示出在不利条件下增强可见图像的能力。然而，RGB-T融合仍然存在交叉模态特征丢失、对环境干扰敏感、生成时间长等问题。产生这些限制的原因是：(1)仅在共享前向网络中难以充分提取特定于模态的信息；(2)忽视恶劣天气条件的干扰；(3)扩散模型的多步去噪过程，增加了时间成本。为了克服这些挑战，我们提出了一种新的RGB-T图像融合条件扩散模型CDMFusion，该模型包含：(1)为融合设计的三分支网络，以更充分地保留信息；(2)多场景自适应特征增强器，在抑制干扰的同时动态增强有价值的特征；(3)一种新的跳跃巡逻机制，无需额外训练即可通过两步去噪实现高质量的生成。此外，还发布了一个新的多场景RGB-T图像数据集和一个多干扰数据集进行综合评价。实验表明，与14种最先进的方法相比，我们的方法在7个数据集上取得了更好的性能。代码和数据集在https://github.com/yangluojie/CDM。

{"title":"Conditional diffusion model for infrared and visible image fusion in open environments with few denoising steps","authors":"Luojie Yang , Chunming Li , Meng Yu , Guangyan Chen , Yufeng Yue","doi":"10.1016/j.sigpro.2025.110437","DOIUrl":"10.1016/j.sigpro.2025.110437","url":null,"abstract":"<div><div>Multi-modality fusion improves perceptual robustness and accuracy by leveraging multi-source sensor data. Current RGB-T fusion methods still falter with adverse illumination and weather. Recent advances in generative methods especially diffusion models have shown the ability to enhance visible images under adverse conditions. However, the fusion of RGB-T still suffer from cross-modal feature loss, sensitivity to environmental interference, and prolonged generation times. These limitations arise due to: (1) difficulties in sufficiently extracting modality-specific information only within shared forward networks; (2) neglecting the interference from adverse weather conditions; (3) the multi-step denoising process in diffusion-based models, which increases temporal cost. To overcome these challenges, we propose a novel conditional diffusion model for RGB-T image fusion, named CDMFusion, which incorporates: (1) a three-branch network designed for fusion to more fully preserve information; (2) a multi-scene adaptive feature enhancer that dynamically enhances valuable features while mitigating interference; (3) a novel skip patrol mechanism enabling high-quality generation via two-step denoising without extra training. Additionally, a new multi-scene RGB-T image dataset and a dataset with multi-interference are released for comprehensive evaluation. Experiments demonstrate our method achieves superior performance across 7 datasets compared to 14 state-of-the-art methods. Code and datasets are at <span><span>https://github.com/yangluojie/CDM</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":49523,"journal":{"name":"Signal Processing","volume":"243 ","pages":"Article 110437"},"PeriodicalIF":3.6,"publicationDate":"2025-12-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145842704","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

VT-BM3D: A collaborative filtering framework with joint optimization of structure awareness and noise characteristics VT-BM3D：结构感知与噪声特性联合优化的协同滤波框架

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

Signal Processing

Pub Date : 2025-12-11 DOI: 10.1016/j.sigpro.2025.110417

Pai Peng, Xinyue Zhang, Yang Cheng, Zhiyu Wu

The classical block-matching and three-dimensional collaborative filtering (BM3D) algorithm employs fixed thresholds for block matching and transform-domain parameters. As a result, it struggles to balance detail preservation and noise suppression, especially under complex image structures or diagonal stripe noise, limiting robustness. To overcome these issues, we propose the Variance–Tensor BM3D (VT-BM3D) framework, which jointly optimizes structure awareness and noise adaptivity while maintaining efficiency and interpretability. First, an adaptive block-matching mechanism is designed using local variance and the structure tensor, enabling dynamic threshold adjustment and significantly improving the capture of fine textures and subtle edges. Second, Canny-based edge protection and texture-region gain modulation are introduced to selectively enhance the recovery of critical visual structures. Third, parameter presets are provided for high-frequency and periodic noise, while the transform domain is dynamically selected between the discrete cosine transform (DCT) and the discrete sine transform (DST) according to local characteristics, further improving adaptivity. Experiments on the Set12 and BSDS300 datasets show that VT-BM3D achieves an average PSNR improvement of 2.04 dB over BM3D, with a maximum gain of 4.19 dB on complex textures, demonstrating superior trade-offs among denoising performance, structural fidelity, and computational efficiency.

经典的块匹配和三维协同滤波（BM3D）算法对块匹配和变换域参数采用固定阈值。因此，它很难平衡细节保存和噪声抑制，特别是在复杂的图像结构或对角条纹噪声，限制了鲁棒性。为了克服这些问题，我们提出了方差张量BM3D （var - tensor BM3D）框架，该框架在保持效率和可解释性的同时，共同优化了结构感知和噪声自适应。首先，设计了一种基于局部方差和结构张量的自适应块匹配机制，实现了阈值的动态调整，显著提高了精细纹理和细微边缘的捕获；其次，引入基于边缘保护和纹理区域增益调制来选择性地增强关键视觉结构的恢复。第三，对高频和周期性噪声进行参数预设，同时根据局部特征在离散余弦变换（DCT）和离散正弦变换（DST）之间动态选择变换域，进一步提高自适应能力。在Set12和BSDS300数据集上的实验表明，VT-BM3D的平均PSNR比BM3D提高了2.04 dB，在复杂纹理上的最大增益为4.19 dB，在去噪性能、结构保真度和计算效率之间取得了更好的平衡。

{"title":"VT-BM3D: A collaborative filtering framework with joint optimization of structure awareness and noise characteristics","authors":"Pai Peng, Xinyue Zhang, Yang Cheng, Zhiyu Wu","doi":"10.1016/j.sigpro.2025.110417","DOIUrl":"10.1016/j.sigpro.2025.110417","url":null,"abstract":"<div><div>The classical block-matching and three-dimensional collaborative filtering (BM3D) algorithm employs fixed thresholds for block matching and transform-domain parameters. As a result, it struggles to balance detail preservation and noise suppression, especially under complex image structures or diagonal stripe noise, limiting robustness. To overcome these issues, we propose the Variance–Tensor BM3D (VT-BM3D) framework, which jointly optimizes structure awareness and noise adaptivity while maintaining efficiency and interpretability. First, an adaptive block-matching mechanism is designed using local variance and the structure tensor, enabling dynamic threshold adjustment and significantly improving the capture of fine textures and subtle edges. Second, Canny-based edge protection and texture-region gain modulation are introduced to selectively enhance the recovery of critical visual structures. Third, parameter presets are provided for high-frequency and periodic noise, while the transform domain is dynamically selected between the discrete cosine transform (DCT) and the discrete sine transform (DST) according to local characteristics, further improving adaptivity. Experiments on the Set12 and BSDS300 datasets show that VT-BM3D achieves an average PSNR improvement of 2.04 dB over BM3D, with a maximum gain of 4.19 dB on complex textures, demonstrating superior trade-offs among denoising performance, structural fidelity, and computational efficiency.</div></div>","PeriodicalId":49523,"journal":{"name":"Signal Processing","volume":"243 ","pages":"Article 110417"},"PeriodicalIF":3.6,"publicationDate":"2025-12-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145802316","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

Polynomial order selection for Savitzky-Golay smoothers via N-fold cross-validation 基于n次交叉验证的Savitzky-Golay平滑多项式阶选择

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

Signal Processing

Pub Date : 2025-12-11 DOI: 10.1016/j.sigpro.2025.110443

Çağatay Candan

Savitzky-Golay (SG) smoothers are noise suppressing filters operating on the principle of projecting noisy input onto the subspace of polynomials. A poorly selected polynomial order results in over- or under-smoothing which shows as either bias or excessive noise at the output. In this study, we apply the N-fold cross-validation technique (also known as leave-one-out cross-validation) for model order selection and show that the inherent analytical structure of the SG filtering problem, mainly its minimum norm formulation, enables an efficient and effective order selection solution. More specifically, a novel connection between the total prediction error and SG-projection spaces is developed to reduce the implementation complexity of cross-validation method. The suggested solution compares favorably with the state-of-the-art Bayesian Information Criterion (BIC) rule in non-asymptotic signal-to-noise ratio (SNR) and sample size regimes. MATLAB codes reproducing the numerical results are provided.

Savitzky-Golay （SG）平滑器是一种噪声抑制滤波器，其工作原理是将噪声输入投影到多项式的子空间上。选择不当的多项式阶导致过平滑或欠平滑，在输出处表现为偏置或过度噪声。在本研究中，我们将N-fold交叉验证技术（也称为留一交叉验证）应用于模型顺序选择，并表明SG滤波问题的固有分析结构，主要是其最小范数公式，能够实现高效和有效的顺序选择解决方案。更具体地说，提出了一种新的预测总误差与sg -投影空间之间的联系，以降低交叉验证方法的实现复杂性。建议的解决方案在非渐近信噪比（SNR）和样本量方面优于最先进的贝叶斯信息准则（BIC）规则。给出了再现数值结果的MATLAB代码。

引用次数: 0

Three-dimensional multi-point embedding digital watermarking scheme based on inter-channel correlation 基于信道间相关的三维多点嵌入数字水印方案

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

Signal Processing

Pub Date : 2025-12-10 DOI: 10.1016/j.sigpro.2025.110447

Shuai Jiao, Yunfei Qiu, Qingtang Su, Ziyang Liu, Chongyu Shi

With the large-scale deployment of 5G networks, the massive data explosion in scenarios has greatly increased the risk of data transmission. To address the issue of real-time interception and tampering during data transmission, this paper proposes a three-dimensional multi-point collaborative embedded digital watermarking scheme based on RGB channel correlation. The energy concentration points are located through Hessenberg decomposition. Three-dimensional joint Quantization Index Modulation (QIM) is implemented at the three positions with the maximum energy concentration. To maintain the invisibility of the watermark, the quantization step size is selected based on RGB channel correlation, and in terms of security, four-dimensional Chen chaotic encryption is introduced, which expands the key space to 2⁴³⁴. Experiments show that when the PSNR reaches a visual fidelity of 40 dB, the average NC value of this scheme exceeds 0.92, achieving a good balance between robustness and invisibility.

随着5G网络的大规模部署，海量数据的场景爆炸大大增加了数据传输的风险。为解决数据传输过程中的实时拦截和篡改问题，提出了一种基于RGB信道相关的三维多点协同嵌入式数字水印方案。通过海森伯格分解确定能量集中点。在能量集中最大的三个位置进行三维联合量化指数调制（QIM）。为了保持水印的不可见性，基于RGB信道相关性选择量化步长，在安全性方面，引入四维陈混沌加密，将密钥空间扩展到2434。实验表明，当PSNR达到40 dB的视觉保真度时，该方案的平均NC值超过0.92，在鲁棒性和不可见性之间取得了很好的平衡。

引用次数: 0

WtCAFNet: A wavelet transform and cross-attention modality-adaptive fusion network for multispectral object detection 基于小波变换和交叉注意模态自适应的多光谱目标检测网络

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

Signal Processing

Pub Date : 2025-12-10 DOI: 10.1016/j.sigpro.2025.110446

Jie Hu , Lu Ni , Bo Peng , Tianrui Li

Multispectral target detection has attracted much attention in recent years because of its excellent performance in low light, smoke and other harsh environments by blending optical and infrared spectral modal information. However, the existing methods still face challenges such as modal feature misalignment, information redundancy and difficult trade-off of complementary features. In this paper, to solve the above problems, we propose a multispectral target detection network that integrates cross-attention mechanism and wavelet convolution. Firstly, a feature fusion module is designed to reduce the impact of pixel-level misalignment, and a lightweight channel attention mechanism is introduced to dynamically adjust the feature weights. At the same time, wavelet convolution is used to extract low-frequency feature information to enhance feature representation ability. Experimental results on FLIR, LLVIP, M3FD and KAIST datasets show that the proposed method outperforms the state-of-the-art methods and is suitable for a variety of practical application scenarios.

近年来，多光谱目标检测因其混合光学和红外光谱模态信息在弱光、烟雾等恶劣环境下的优异性能而备受关注。然而，现有方法仍然面临着模态特征不匹配、信息冗余和互补特征权衡困难等问题。为了解决上述问题，本文提出了一种融合交叉注意机制和小波卷积的多光谱目标检测网络。首先，设计特征融合模块降低像素级不匹配的影响，引入轻量级通道关注机制动态调整特征权重；同时，利用小波卷积提取低频特征信息，增强特征表示能力。在FLIR、LLVIP、M3FD和KAIST数据集上的实验结果表明，该方法优于现有方法，适用于多种实际应用场景。

引用次数: 0

首页上一页

下一页尾页

类型

全部化学•材料生命科学医学物理工程技术环境•农林材料科学地球科学法学管理学化学环境科学与生态学计算机科学教育学经济学农林科学人文科学生物学数学物理与天体物理心理学综合性期刊其他工业工程理学历史学农学文学信息工程

数据库

全部 ACS Publications Elsevier ieeexplore Springer The Royal Society of Chemistry Wiley

期刊

Signal Processing

全部 Acc. Chem. Res. ACS Applied Bio Materials ACS Appl. Electron. Mater. ACS Appl. Energy Mater. ACS Appl. Mater. Interfaces ACS Appl. Nano Mater. ACS Appl. Polym. Mater. ACS BIOMATER-SCI ENG ACS Catal. ACS Cent. Sci. ACS Chem. Biol. ACS Chemical Health & Safety ACS Chem. Neurosci. ACS Comb. Sci. ACS Earth Space Chem. ACS Energy Lett. ACS Infect. Dis. ACS Macro Lett. ACS Mater. Lett. ACS Med. Chem. Lett. ACS Nano ACS Omega ACS Photonics ACS Sens. ACS Sustainable Chem. Eng. ACS Synth. Biol. Anal. Chem. BIOCHEMISTRY-US Bioconjugate Chem. BIOMACROMOLECULES Chem. Res. Toxicol. Chem. Rev. Chem. Mater. CRYST GROWTH DES ENERG FUEL Environ. Sci. Technol. Environ. Sci. Technol. Lett. Eur. J. Inorg. Chem. IND ENG CHEM RES Inorg. Chem. J. Agric. Food. Chem. J. Chem. Eng. Data J. Chem. Educ. J. Chem. Inf. Model. J. Chem. Theory Comput. J. Med. Chem. J. Nat. Prod. J PROTEOME RES J. Am. Chem. Soc. LANGMUIR MACROMOLECULES Mol. Pharmaceutics Nano Lett. Org. Lett. ORG PROCESS RES DEV ORGANOMETALLICS J. Org. Chem. J. Phys. Chem. J. Phys. Chem. A J. Phys. Chem. B J. Phys. Chem. C J. Phys. Chem. Lett. Analyst Anal. Methods Biomater. Sci. Catal. Sci. Technol. Chem. Commun. Chem. Soc. Rev. CHEM EDUC RES PRACT CRYSTENGCOMM Dalton Trans. Energy Environ. Sci. ENVIRON SCI-NANO ENVIRON SCI-PROC IMP ENVIRON SCI-WAT RES Faraday Discuss. Food Funct. Green Chem. Inorg. Chem. Front. Integr. Biol. J. Anal. At. Spectrom. J. Mater. Chem. A J. Mater. Chem. B J. Mater. Chem. C Lab Chip Mater. Chem. Front. Mater. Horiz. MEDCHEMCOMM Metallomics Mol. Biosyst. Mol. Syst. Des. Eng. Nanoscale Nanoscale Horiz. Nat. Prod. Rep. New J. Chem. Org. Biomol. Chem. Org. Chem. Front. PHOTOCH PHOTOBIO SCI PCCP Polym. Chem.

﹀