Signal Processing最新文献_第3页

LatentDark: Reflectance guided latent diffusion model for low-light image enhancement LatentDark：用于弱光图像增强的反射引导潜扩散模型

IF 3.4 2区工程技术

Signal Processing Pub Date : 2025-06-09 DOI: 10.1016/j.sigpro.2025.110125

Renzhi Hu , Ting Luo , Gangyi Jiang , Leiming Liu , Yeyao Chen , Haiyong Xu , Zhouyan He

{"title":"LatentDark: Reflectance guided latent diffusion model for low-light image enhancement","authors":"Renzhi Hu , Ting Luo , Gangyi Jiang , Leiming Liu , Yeyao Chen , Haiyong Xu , Zhouyan He","doi":"10.1016/j.sigpro.2025.110125","DOIUrl":"10.1016/j.sigpro.2025.110125","url":null,"abstract":"<div><div>Existing diffusion-based low-light image enhancement (LIE) models operate on original resolution images, resulting in low recovery efficiency and limiting their practical applications. To reduce computational resources and inference time, Latent diffusion models (LDM) utilize an autoencoder (AE) to compress images into a low-resolution latent space. However, this dimensionality reduction inevitably leads to the loss of texture and color information due to the limited capacity of the latent representation. To efficiently recover image texture details and color, we propose a reflectance guided LDM, namely LatentDark. Specifically, the reflectance image decomposed by Retinex is used to guide the reconstruction, thereby compensating for the texture and color loss caused by the low-dimensional encoding. During the diffusion process, the reflectance features are embedded in the latent space to guide denoising, and a multi-conditional attention fusion module (MAFM) is designed to fully mine various essential features from different conditions. Additionally, we propose a lightweight noise estimation network (NENet) that can quickly and accurately predict noise, further improving inference speed. Extensive experiments on publicly available datasets demonstrate that LatentDark excels in both quantitative and visual quality, and significantly outperforms other diffusion-based LIE models in terms of inference time. The results and code will be made publicly accessible at <span><span>https://github.com/RenzhiHu111/LatentDark</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":49523,"journal":{"name":"Signal Processing","volume":"238 ","pages":"Article 110125"},"PeriodicalIF":3.4,"publicationDate":"2025-06-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144279267","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

Remote state estimation with stochastic event-triggered Kalman filtering with non-Gaussian environment 非高斯环境下随机事件触发卡尔曼滤波的远程状态估计

IF 3.4 2区工程技术

Signal Processing Pub Date : 2025-06-06 DOI: 10.1016/j.sigpro.2025.110121

Junwen Deng , Kunpeng Zhang , Gang Wang , Bolin Yang , Meng Fu , Jiacheng He

引用次数: 0

Document image forgery detection and localization in desensitization scenarios 脱敏场景下文档图像伪造检测与定位

IF 3.4 2区工程技术

Signal Processing Pub Date : 2025-06-06 DOI: 10.1016/j.sigpro.2025.110123

Weixiang Li, Bin Li, Kengtao Zheng, Songze Li, Haodong Li

{"title":"Document image forgery detection and localization in desensitization scenarios","authors":"Weixiang Li, Bin Li, Kengtao Zheng, Songze Li, Haodong Li","doi":"10.1016/j.sigpro.2025.110123","DOIUrl":"10.1016/j.sigpro.2025.110123","url":null,"abstract":"<div><div>Document images are widely used in e-commerce, and some privacy information contained in them may be desensitized before circulation. Since innocuous desensitization is rather different from malicious tampering in both motivation and appearance, it results in a new forensic scenario, in which reliable forgery detection and localization is needed when desensitization artifacts present. In this paper, we address the issue for the first time by proposing DCLNet (Desensitization involved Contrastive Learning based forensic Network), to improve the performance of pixel-level tampering localization and image-level forgery detection. DCLNet is built upon a ConvNeXt-based encoder–decoder network with a global context attention module, enabling it to learn effective features from multi-scales. To tackle the difficulty of learning weak tampering traces without interference from strong desensitization artifacts, we design a contrastive learning module to effectively differentiate between the two kinds of manipulations. Additionally, we construct a document image dataset that considers various document types and contains both tampering and desensitization manipulations, providing sufficient data for training and evaluation. Extensive experimental results demonstrate that DCLNet outperforms existing methods for the new task, and exhibits good robustness to post-processing and better adaptability to other sources of document images.</div></div>","PeriodicalId":49523,"journal":{"name":"Signal Processing","volume":"238 ","pages":"Article 110123"},"PeriodicalIF":3.4,"publicationDate":"2025-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144271636","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

Joint design of transmit and receive array window functions via polynomial factorization 利用多项式分解法联合设计发射与接收阵列窗函数

IF 3.4 2区工程技术

Signal Processing Pub Date : 2025-06-05 DOI: 10.1016/j.sigpro.2025.110124

Håvard Kjellmo Arnestad, Andreas Austeng, Sven Peter Näsholm

{"title":"Joint design of transmit and receive array window functions via polynomial factorization","authors":"Håvard Kjellmo Arnestad, Andreas Austeng, Sven Peter Näsholm","doi":"10.1016/j.sigpro.2025.110124","DOIUrl":"10.1016/j.sigpro.2025.110124","url":null,"abstract":"<div><div>Window functions are commonly used to balance the mainlobe width and sidelobe levels in beamforming applications. Traditionally, windows are selected independently for the transmit (Tx) and receive (Rx) side. However, in active systems like radar, medical ultrasound, and sonar, the same array may operate in both modes, and the convolution of the Tx and Rx windows determines the effective two-way response. Therefore, without jointly designing the Tx/Rx windows, the mainlobe and sidelobe characteristics do not fully benefit from the possibilities of a consolidated two-way approach. This paper presents a framework for jointly designing Tx/Rx window functions by factorization of any desired two-way effective aperture into separate Tx/Rx windows. To guide this factorization, we introduce the white noise gain product (WNGP), a metric quantifying the combined spatial filtering efficiency of the Tx/Rx pair. We then propose a root-allocation strategy for the factorization that maximizes this metric, enabling effective control over the two-way beampattern. Additionally, we derive a recursive formula for the optimal Tx/Rx windows that yield a uniform effective aperture. The approach is validated through simulations, showing improved sidelobe control and greater design flexibility compared to conventional windowing techniques.</div></div>","PeriodicalId":49523,"journal":{"name":"Signal Processing","volume":"238 ","pages":"Article 110124"},"PeriodicalIF":3.4,"publicationDate":"2025-06-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144255287","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

Wideband MIMO radar beampattern shaping in spectrally dense environments 频谱密集环境下宽带MIMO雷达波束成形

IF 3.4 2区工程技术

Signal Processing Pub Date : 2025-06-05 DOI: 10.1016/j.sigpro.2025.110122

Dongxu An , Hua Wang , Jinfeng Hu , Xin Tai , Xinsheng Peng , Kai Zhong , Yongfeng Zuo , Huiyong Li , Fulvio Gini

{"title":"Wideband MIMO radar beampattern shaping in spectrally dense environments","authors":"Dongxu An , Hua Wang , Jinfeng Hu , Xin Tai , Xinsheng Peng , Kai Zhong , Yongfeng Zuo , Huiyong Li , Fulvio Gini","doi":"10.1016/j.sigpro.2025.110122","DOIUrl":"10.1016/j.sigpro.2025.110122","url":null,"abstract":"<div><div>Wideband MIMO radar beampattern shaping with Constant Modulus Constraints (CMCs) in spectrally dense environments is critical for future 6G networked sensing technology. Existing methods minimize the weighted function of wideband MIMO radar beampattern matching Mean Square Error (MSE) and the Energy Spectral Density (ESD) of Spatial Spectral Nulling (SSN) region; however, achieving precise ESD control remains a challenge. To address this, we minimize the beampattern matching MSE under CMCs and precise SSN Constraints (SSNCs). The non-convex nature of the CMCs and multiple SSNCs lead to a non-convex Quadratic-Constrained Quadratic Programming (QCQP) problem. To solve the problem, we propose a novel Manifold-Based Exact Penalty (MBEP) method. First, we construct the Complex Circular Manifold (CCM) to satisfy the CMCs and reformulate the SSNCs as an exact penalty function, thereby transforming the problem into an unconstrained optimization problem on the CCM. Subsequently, a Simplified Quasi-Newton (SQN) method is developed to optimize the problem on the CCM. Finally, the penalty factor is adaptively updated to improve the optimization process. Compared with existing methods: 1) the proposed method achieves precise control of the ESD level in the SSN region; and 2) the ESD in the SSN region is reduced by 8.8 dB, while the beampattern matching MSE is decreased by 0.02 dB.</div></div>","PeriodicalId":49523,"journal":{"name":"Signal Processing","volume":"238 ","pages":"Article 110122"},"PeriodicalIF":3.4,"publicationDate":"2025-06-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144291645","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

Distributed multisensor adaptive PMB filter with inaccurate measurement noise covariances 测量噪声协方差不准确的分布式多传感器自适应PMB滤波器

IF 3.4 2区工程技术

Signal Processing Pub Date : 2025-06-04 DOI: 10.1016/j.sigpro.2025.110117

Xiangfei Zheng, Yujie Zhang, Hongwei Li

{"title":"Distributed multisensor adaptive PMB filter with inaccurate measurement noise covariances","authors":"Xiangfei Zheng, Yujie Zhang, Hongwei Li","doi":"10.1016/j.sigpro.2025.110117","DOIUrl":"10.1016/j.sigpro.2025.110117","url":null,"abstract":"<div><div>In most multi-target tracking (MTT) scenarios, the prior information about the statistical characteristics of the measurement noise usually needs to be more accurate and pre-trained. However, in practice, this is often difficult to obtain accurately, especially when additive noise and multiplicative noise exist at the same time. In this case, the adaptive estimation of the measurement noise covariances is quite important for MTT. In this paper, an adaptive Poisson multi-Bernoulli (APMB) filter based on the random finite sets framework and variational Bayesian inference is proposed, which considers additive noise and multiplicative noise as a whole with a unified covariance. Moreover, the Gaussian inverse Wishart (GIW) implementation of the proposed APMB filter is given, where the Gaussian distribution describes the kinematics of the target, the IW distribution describes the covariance of the measurement noise, and the mixing of the GIW components in the Bernoulli components is calculated on average according to the weighted Kullback–Leibler. Furthermore, the proposed APMB filter is applied to a distributed multisensor, which can deal with inaccurate and inconsistent measurement noise covariances between different sensors. Finally, simulation results show that the proposed APMB filter can estimate the target effectively when the measurement noise covariances are inaccurate.</div></div>","PeriodicalId":49523,"journal":{"name":"Signal Processing","volume":"238 ","pages":"Article 110117"},"PeriodicalIF":3.4,"publicationDate":"2025-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144255288","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

Learning transferable representations by topic guided graph adversarial network 用主题导向图对抗网络学习可转移表征

IF 3.4 2区工程技术

Signal Processing Pub Date : 2025-06-04 DOI: 10.1016/j.sigpro.2025.110118

Zhengjue Wang , Zhihui Xin , Chiyu Chen , Hao Zhang , Yuhao Huang , Yunsong Li , Hongwei Liu , Bo Chen

{"title":"Learning transferable representations by topic guided graph adversarial network","authors":"Zhengjue Wang , Zhihui Xin , Chiyu Chen , Hao Zhang , Yuhao Huang , Yunsong Li , Hongwei Liu , Bo Chen","doi":"10.1016/j.sigpro.2025.110118","DOIUrl":"10.1016/j.sigpro.2025.110118","url":null,"abstract":"<div><div>To reduce the need for costly labeled data in a target domain, it is desirable to learn transferable representations from a related source domain, where it is feasible to obtain labeled data. Facing distribution shift between different domains, one often aligns the marginal feature distributions by performing feature-level adversarial learning. Despite the recent success, existing approaches often ignore the consistency between feature and label, which is a challenging problem since there is no supervision in the target domain. Hence, we are motivated to transfer the discriminative information from the source to the target domain to realize better domain alignment. To this end, we propose a topic-guided graph adversarial network (TGAN), composed of a graph constructor, a graph feature extractor, a domain discriminator, and a classifier. Specifically, to learn a graph describing the relational structure among samples from different domains, we propose a semantic disentangled topic model to extract domain-shared and domain-specific topics, so that the graph edges can be defined by the sample similarities in a domain-shared semantic space. Then, TGAN aggregates the discriminative characteristics of source nodes and propagates them to the target nodes by attentive message passing through the graph, with the final node embeddings used for adversarial learning between source and target domains. TGAN has achieved the state-of-the-art performance on sentiment classification and clinical risk prediction tasks. Moreover, the discovered domain-invariant discriminative topics show interpretable meanings, which is beneficial to understanding the prediction results, especially for biomedical researches.</div></div>","PeriodicalId":49523,"journal":{"name":"Signal Processing","volume":"238 ","pages":"Article 110118"},"PeriodicalIF":3.4,"publicationDate":"2025-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144271635","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

Cramér–Rao Lower Bound of adaptive filtering algorithms for acoustic echo cancellation 声学回波抵消自适应滤波算法的cram<s:1> - rao下界

IF 3.4 2区工程技术

Signal Processing Pub Date : 2025-06-03 DOI: 10.1016/j.sigpro.2025.110111

Zongsheng Zheng , Ziyuan Shao , Yi Yu , Lu Lu , Shilin Gao

{"title":"Cramér–Rao Lower Bound of adaptive filtering algorithms for acoustic echo cancellation","authors":"Zongsheng Zheng , Ziyuan Shao , Yi Yu , Lu Lu , Shilin Gao","doi":"10.1016/j.sigpro.2025.110111","DOIUrl":"10.1016/j.sigpro.2025.110111","url":null,"abstract":"<div><div>Numerous adaptive filtering algorithms have been proposed for acoustic echo cancellation. However, whether the performance of the algorithms approaches the optimal performance or if there has been intentionally overstated remains challenging to evaluate. Fortunately, the Cramér–Rao Lower Bound (CRLB) provides a theoretical minimum variance for any unbiased estimator under given observational data and statistical models. This paper derives the CRLB of adaptive filtering algorithms for acoustic echo cancellation (AEC), in which the generalized Gaussian distribution (GGD) is utilized to model the Gaussian/non-Gaussian background noises. To accelerate the CRLB calculation process, the recursive resolution of the CRLB is presented by using the matrix inversion lemma, and the computational complexity is also analyzed. The derivation results indicate that CRLB for AEC model depends on the acoustic input (i.e., speaker’s voice) and the statistical properties of GGD noise but is unaffected by the channel sparsity. The CRLB derived in this paper can serve as a benchmark to evaluate whether the performance of the adaptive filtering algorithms is optimal and to exclude some adaptive filtering algorithms that deliberately exaggerate their performance.</div></div>","PeriodicalId":49523,"journal":{"name":"Signal Processing","volume":"238 ","pages":"Article 110111"},"PeriodicalIF":3.4,"publicationDate":"2025-06-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144231523","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

Closed-form expression for resolution limit of direction-of-arrival estimation in co-prime array 共素数阵列到达方向估计分辨率极限的封闭表达式

IF 3.4 2区工程技术

Signal Processing Pub Date : 2025-06-03 DOI: 10.1016/j.sigpro.2025.110112

Han Zhang , Jingjing Pan , Dazhuan Xu , Xiaofei Zhang , Xudong Dong

{"title":"Closed-form expression for resolution limit of direction-of-arrival estimation in co-prime array","authors":"Han Zhang , Jingjing Pan , Dazhuan Xu , Xiaofei Zhang , Xudong Dong","doi":"10.1016/j.sigpro.2025.110112","DOIUrl":"10.1016/j.sigpro.2025.110112","url":null,"abstract":"<div><div>Utilizing co-prime linear arrays (CLA) in place of uniform linear arrays can greatly improve the direction-of-arrival (DOA) resolution with the same number of elements. However, the explicit DOA resolution limit (DRL) of the CLA is unavailable and the resolution gain has not been investigated sufficiently. In this work, Shannon’s information theory is utilized to establish the closed-form expression of the DRL of the CLA. For complex Gaussian sources, we derive the scattering information of duo-source whose amplitudes are the same. A critical state is picked in which the scattering information’s quadrature part equals 1 bit, and the DOA separation is determined to be the DRL. The explicit DRL is then obtained by a Taylor expansion, which is algorithm-independent and can be applied to all signal-to-noise ratios. The expression illustrates that the DRL is approximately inversely proportional to the direction cosine, the root-mean-square aperture width, and the square root of the signal-to-noise ratio. In addition, the quantitative relationship between the number of elements, sparse array aperture, and the optimal resolution limits of three kinds of common-used sparse arrays is obtained, which is of practical significance to the sparse array design.</div></div>","PeriodicalId":49523,"journal":{"name":"Signal Processing","volume":"238 ","pages":"Article 110112"},"PeriodicalIF":3.4,"publicationDate":"2025-06-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144212546","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

Tensor-based higher-order multivariate singular spectrum analysis and applications to multichannel biomedical signal analysis 基于张量的高阶多元奇异谱分析及其在多通道生物医学信号分析中的应用

IF 3.4 2区工程技术

Signal Processing Pub Date : 2025-06-02 DOI: 10.1016/j.sigpro.2025.110113

Thanh Trung Le , Karim Abed-Meraim , Nguyen Linh Trung , Philippe Ravier , Olivier Buttelli , Ales Holobar

{"title":"Tensor-based higher-order multivariate singular spectrum analysis and applications to multichannel biomedical signal analysis","authors":"Thanh Trung Le , Karim Abed-Meraim , Nguyen Linh Trung , Philippe Ravier , Olivier Buttelli , Ales Holobar","doi":"10.1016/j.sigpro.2025.110113","DOIUrl":"10.1016/j.sigpro.2025.110113","url":null,"abstract":"<div><div>Singular spectrum analysis (SSA) is a nonparametric spectral estimation method that decomposes time series signals into interpretable components. With the rise of big time series, the demand for effective and scalable SSA techniques has become increasingly urgent. In this paper, we propose a novel multiway extension of SSA, called higher-order multivariate SSA (HO-MSSA), specifically designed for multivariate and multichannel time series signal analysis via tensor decomposition. HO-MSSA utilizes time-delay embedding and tensor singular value decomposition to transform multichannel time series signals into trajectory tensors, which are then decomposed into elementary components in the Fourier domain, rather than the time domain as in traditional SSA methods. These components are grouped into disjoint subsets using spectral clustering, enabling the reconstruction of the underlying source signals. Experimental results demonstrate that HO-MSSA outperforms state-of-the-art SSA methods in various biomedical applications, including electromyography (EMG), electrocardiography (ECG), and electroencephalogram (EEG) signals.</div></div>","PeriodicalId":49523,"journal":{"name":"Signal Processing","volume":"238 ","pages":"Article 110113"},"PeriodicalIF":3.4,"publicationDate":"2025-06-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144203924","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