检索结果-内蒙古大学图书馆

Enhanced Detection of Narrow-Band Leakage Gas Infrared Image Based on Bilateral filtering and U-Net

IEEE ACCESS 2025年 13卷 55559-55575页

作者： Dong, Kangjun Jin, Weiqi Wang, Minghe Wang, Xia Xu, Shiwei Beijing Inst Technol Sch Opt & Photon MOE Key Lab Optoelect Imaging Technol & Syst Beijing 100081 Peoples R China

Uncooled infrared image detection methods with a narrow-band filter have the advantages of low cost and can efficiently distinguish gas types;however, they also introduce nonuniform patterns to the image and reduce the target radiation flux and signal-to-noise ratio (SNR), resulting in low gas extraction accuracy. In this study, first, a temperature control device is designed in a part containing a filter and a detector array to effectively suppress dark current noise, which is positively correlated with temperature and eliminates nonuniform patterns, such as the lid effect, using the two-point correction method. Second, according to the infrared image characteristics of the leaked gas, a leakage gas infrared image enhancement algorithm based on bilateral filtering is developed. This algorithm considerably reduces the time frame noise of image sequences, improves the contrast between the gas plume region and background, and improves the SNR. Next, based on an improved fuzzy Gaussian background model, U-Net is used to separate the gas and non-gas foreground. Experimental results demonstrate that the proposed method exhibits high gas plume detection accuracy.

关键词： Image enhancement Dynamic range Feature extraction Signal to noise ratio Low-pass filters Information filters Histograms Optical filters filtering algorithms Visualization Bilateral filter gas leak detection image enhancement infrared imaging narrow band filter

来源：评论

学校读者我要写书评

暂无评论

Seizure Detection Algorithm Based on Hybrid Morphological filtering Optimization Multiscale Amplitude Integrated EEG

引用

IEEE SENSORS JOURNAL 2025年第8期25卷 13529-13540页

作者： Dong, Yongping Wu, Duanpo Jiang, Lurong Dong, Fang Liu, Junbiao Vidal, Pierre-Paul Cao, Jiuwen Wang, Danping Hangzhou Dianzi Univ Sch Commun Engn Hangzhou 310018 Peoples R China Hangzhou Dianzi Univ Artificial Intelligence Inst Hangzhou 310018 Zhejiang Peoples R China Zhejiang Sci Tech Univ Sch Informat Sci & Technol Hangzhou 310018 Peoples R China Hangzhou City Univ Sch Informat & Elect Engn Hangzhou 310015 Peoples R China French Natl Ctr Sci Res F-75016 Paris France Hangzhou Dianzi Univ Machine Learning & I Hlth Int Cooperat Base Zhejia Hangzhou 310018 Zhejiang Peoples R China Univ Paris Cite Plateforme Etud Sensorimotr PES BioMedTech Facil F-75270 Paris France

Automatic seizure detection algorithms have significant implications for improving patients' quality of life. However, the absence of multiscale analysis in these algorithms regarding electroencephalography (EEG) signals may lead to overlooking crucial seizure features. To address this limitation, our study aims to develop a novel seizure detection algorithm based on hybrid morphological filtering optimization multiscale amplitude integrated EEG (aEEG). First, multichannel EEG signals, collected by EEG sensors, are preprocessed with asymmetric filtering, rectification, envelope detection, multiscale time compression and signal smoothing to obtain multiscale aEEG. Second, morphological filter (MF), multiscale MF (MSMF), and average combination difference MF (ACDMF) are hybridized to filter, analyze, and reconstruct the signals, thereby emphasizing the seizures. Thereafter, optimal aEEG signal is selected using the channel and time compression scale (TCS) joint optimization algorithm based on the kappa (Ka), and this signal is utilized for seizure detection. Finally, it is evaluated on the CHB-MIT database using multiple validation methods, achieving satisfactory experimental results. The results indicate that the model attains accuracy (ACC), specificity (SPE), sensitivity (SEN), F1 score ( F1 ), and Ka of 97.60%, 96.56%, 98.63%, 97.60%, and 95.19%, respectively.

关键词： Electroencephalography filtering Feature extraction Sensors filtering algorithms Electronic mail Detection algorithms Optimization Smoothing methods filtering theory Electroencephalography (EEG) sensor signal hybrid morphology filter multiscale time compression optimal amplitude integrated EEG (aEEG) signal seizure detection

来源：评论

学校读者我要写书评

暂无评论

Improvement of Deep Learning Models Using Retinal Filter: A Systematic Evaluation of the Effect of Gaussian filtering With a Focus on Industrial Inspection Data

引用

IEEE ACCESS 2025年 13卷 43201-43217页

作者： Buettner, Ricardo Mai, Christopher Penava, Pascal Univ Fed Armed Forces Hamburg Helmut Schmidt Univ Chair Hybrid Intelligence D-22043 Hamburg Germany

The range of products manufactured in the industry is vast, as are the anomalies that may occur during the manufacturing process. Some companies have initiated the automation of the quality control process and the implementation of modern deep learning methodologies to promptly identify defective objects. In addition to architectural optimizations within the deep learning model, the utilization of pre-processing filters can also be employed to help improve recognition performance. The Gaussian filter serves as an illustrative example, exhibiting functionality that parallels aspects of the human visual processing system. Although some authors have successfully applied this filter, a comprehensive study assessing its actual impact on a large number of objects and the extent to which various performance indicators are affected remains pending, which is the primary objective of this study. To this end, a variety of architectural approaches are employed, including Xception, InceptionV3, ResNet50V2, VGG19, and VGG16. In addition to hyperparameter tuning, all architectures utilize transfer learning. The results demonstrate that the implementation of an innovative Gaussian filter approach enhances the balanced accuracy in at least one architecture for 13 out of 15 product categories. Furthermore, the filtering approach positively influences various performance indicators across the majority of categories. It can be concluded that the Gaussian filter is often an effective technique for enhancing model performance across various product categories, making it a valuable and efficient tool for industrial defect detection in quality control applications. This study provides an overview of the results achieved and other key performance indicators for all the models used.

关键词： Quality control Deep learning Accuracy Optical filters Defect detection Gaussian distribution filtering algorithms Visualization Retina Transfer learning Gaussian filter defect detection transfer learning pre-trained architectures

来源：评论

学校读者我要写书评

暂无评论

Distributed Consensus-Based filtering Against False Data Injection Attacks

引用

IEEE TRANSACTIONS ON NETWORK SCIENCE AND ENGINEERING 2025年第1期12卷 110-121页

作者： Yang, Yuhang Gao, Xiangzhou Song, Shenmin Li, Zhiqiang Harbin Inst Technol Ctr Control Theory & Guidance Technol Harbin 150001 Peoples R China China Elect Technol Grp Corp 32nd Res Inst Shanghai 201808 Peoples R China

Existing distributed state estimation algorithms usually show satisfactory performance when dealing with data bias caused by network-induced phenomena. However, the security characteristics of these algorithms are often significantly affected by more complex and severe network attacks. Specifically, due to the lack of dynamic adaptability and abnormal data detection ability of the estimator, the estimator may deteriorate significantly or even diverge, which poses a serious threat to the stability and reliability of the system. To remedy this issue, we propose a distributed estimation algorithm based on the classical Kalman consensus filter framework. The accuracy of the estimator is significantly improved by utilizing the innovation of neighbor nodes. Furthermore, we construct an adaptive weight allocation mechanism based on the principle of minimizing the estimation error variance according to the possible accuracy differences between different estimators. This mechanism can evaluate the data accuracy of each node, and dynamically adjust its weight accordingly. Subsenquently, an event-triggered detector with random thresholds is designed to enhance the anti-attack ability of the estimator. The detector can monitor the data flow in the network in real time, and identify the potential abnormal or attack behavior by setting dynamic thresholds. Once abnormal data is detected, the detector can immediately trigger corresponding countermeasures to block the propagation path of erroneous data and protect the safe and stable operation of the system. Simulation results are employed to validate the effectiveness of the proposed method.

关键词： State estimation Accuracy Technological innovation Kalman filters Distributed databases Heuristic algorithms Detectors Estimation error filtering algorithms Convergence False data injection attack wireless sensor networks Kalman consensus filter state estimation

来源：评论

学校读者我要写书评

暂无评论

Error-Mask-Adaptive Dynamic filtering for Image Inpainting

引用

IEEE ACCESS 2025年 13卷 18403-18417页

作者： Ko, Keunsoo Woo, Seunggyun Kim, Chang-Su Catholic Univ Korea Dept Artificial Intelligence Bucheon 14662 South Korea Korea Univ Sch Elect Engn Seoul 02841 South Korea

A novel error-mask-adaptive dynamic filtering (EMDF) algorithm is proposed in this paper, which uses a continuous error mask to inpaint an image adaptively and faithfully. In an EMDF layer, we determine a spatially varying filter adaptively according to an error mask and perform separable dynamic filtering. Meanwhile, we update the error mask by modeling the error propagation during the filtering. Through several EMDF layers, we predict an inpainting result. Finally, we refine it to reconstruct a more faithful image. Experimental results on diverse datasets show that the proposed EMDF algorithm outperforms existing inpainting algorithms significantly. The source codes are available at https://***/keunsoo-ko/EMDF.

关键词： filtering Convolution Image reconstruction Heuristic algorithms filtering algorithms Adaptive filters Adaptation models Prediction algorithms Image restoration Visualization Image inpainting dynamic filtering mask-adaptive filtering convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

DGLG: A Novel Deep Generalized Legendre-Galerkin Approach to Optimal filtering Problem

引用

IEEE TRANSACTIONS ON AUTOMATIC CONTROL 2025年第4期70卷 2584-2590页

作者： Shi, Ji Jiao, Xiaopei Yau, Stephen S. -T. Capital Normal Univ Acad Multidisciplinary Studies Beijing 100048 Peoples R China Univ Twente Dept Appl Math Enschede Netherlands Beijing Inst Math Sci & Applicat Beijing 101400 Peoples R China Tsinghua Univ Dept Math Sci Beijing 100084 Peoples R China

The optimal filtering problem for general nonlinear and continuous state-observation systems attracts lots of attention in the control theory. The essence of optimal filtering requires solving the Duncan-Mortensen-Zakai (DMZ) equation in a computationally feasible way. Under the pioneering work of Yau-Yau filtering, the DMZ equation is reduced to a pathwise computation of a forward Kolmogorov equation with time-varying initial conditions, which is very challenging. To overcome the computational difficulty, in this article, we proposed a new efficient filtering algorithm consisting of a forward Kolmogorov equation solver based on a physics-informed neural network and a probability density approximator based on generalized Legendre polynomials. By utilizing the advanced deep learning method and classical Galerkin approximation, our developed algorithm not only maintains the high accuracy of the spectral method but also removes massive computational loads in the offline part. Furthermore, the convergence of our method is proved. Numerical experiments have been carried out to verify the feasibility of the new method. Regarding accuracy and efficacy, the newly proposed deep generalized Legendre-Galerkin algorithm outperforms other popular suboptimal methods including the extended Kalman filter and particle filter.

关键词： filtering algorithms filtering filtering theory Kalman filters Approximation algorithms Real-time systems Estimation Convergence Probability density function Polynomials filtering generalized Legendre polynomial (GLP) neural networks nonlinear systems

来源：评论

学校读者我要写书评

暂无评论

Distributed Diffusion Cubature Kalman filtering Based on Data Compression and Fast Covariance Intersection

引用

IEEE SENSORS JOURNAL 2024年第19期24卷 31145-31156页

作者： Liu, Jingang Yang, Yuhang Zhang, Zeqi Song, Shenmin Harbin Inst Technol Ctr Control Theory & Guidance Technol Harbin 150001 Peoples R China Harbin Engn Univ Coll Power & Energy Engn Harbin 150001 Peoples R China

In this article, a diffusion cubature Kalman filtering based on data compression and fast covariance intersection (CDCKF-FCI) is proposed for nonlinear sensor networks. The weighted measurement fusion algorithm is applied to compress the measurements of the sensor and its neighbor nodes to obtain a compressed measurement. Based on the compressed measurement and cubature Kalman filter (CKF), a local estimate is derived for each sensor. It does not involve the computation and exchange of pseudo measurement matrices, avoiding the loss of accuracy with statistical linearization and the communication burden between nodes. Subsequently, this approach is proved to be numerically equivalent to centralized fusion. Considering that the correlation of nodes is unknown or unavailable, a fast covariance intersection (CI) fusion algorithm is used for diffusion fusion. It fully considers the estimation errors of the nodes at different time and does not require complex optimization process. Finally, it is proved that the mean performance and the estimation error are exponentially bounded in mean square under some assumptions. A simulation example shows the effectiveness of the proposed algorithm.

关键词： Sensors Kalman filters filtering Data compression Accuracy Loss measurement filtering algorithms Covariance intersection (CI) data compression diffusion filtering

来源：评论

学校读者我要写书评

暂无评论

USDnet: Unsupervised Speech Dereverberation via Neural Forward filtering

引用

IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING 2024年 32卷 3882-3895页

作者： Wang, Zhong-Qiu Southern Univ Sci & Technol Dept Comp Sci & Engn Shenzhen 518055 Peoples R China

In reverberant conditions with a single speaker, each far-field microphone records a reverberant version of the same speaker signal at a different location. In over-determined conditions, where there are multiple microphones but only one speaker, each recorded mixture signal can be leveraged as a constraint to narrow down the solutions to target anechoic speech and thereby reduce reverberation. Equipped with this insight, we propose USDnet, a novel deep neural network (DNN) approach for unsupervised speech dereverberation (USD). At each training step, we first feed an input mixture to USDnet to produce an estimate for target speech, and then linearly filter the DNN estimate to approximate the multi-microphone mixture so that the constraint can be satisfied at each microphone, thereby regularizing the DNN estimate to approximate target anechoic speech. The linear filter can be estimated based on the mixture and DNN estimate via neural forward filtering algorithms such as forward convolutive prediction. We show that this novel methodology can promote unsupervised dereverberation of single-source reverberant speech.

关键词： Training Microphones Mathematical models Reverberation filtering algorithms Nonlinear filters Noise measurement Unsupervised neural speech dereverberation (USD)

来源：评论

学校读者我要写书评

暂无评论

Minimum Total Quaternion Error Entropy filtering With Fiducial Points Against Asymmetric Noise

引用

IEEE SIGNAL PROCESSING LETTERS 2025年 32卷 306-310页

作者： Lin, Dongyuan Cai, Peng Chen, Xiaofeng Zheng, Yunfei Wang, Shiyuan Southwest Univ Coll Elect & Informat Engn Chongqing 400715 Peoples R China Chongqing Jiaotong Univ Dept Math Chongqing 400074 Peoples R China

Quaternion adaptive filters (QAFs) are extensively used in processing three- or four-dimensional signals effectively. However, their performance can significantly deteriorate or even diverge when system inputs and outputs are contaminated by complex noises. Therefore, this letter addresses the issue of parameter estimation in the quaternion errors-in-variables (QEIV) in asymmetric noise. First, a novel robust criterion, called improved quaternion minimum error entropy criterion with fiducial points (IQMEEF), is constructed. Then, a minimum total quaternion error entropy algorithm with fiducial points (MTQEEF) is proposed by integrating the IQMEEF criterion with the total least squares (TLS) method, leveraging stochastic gradient and quaternion generalized Hamilton-real (GHR) calculus theory. Finally, simulations validate the superior performance of MTQEEF in the QEIV model under asymmetric noise environments.

关键词： Quaternions Noise Entropy Signal processing algorithms Adaptation models Vectors Parameter estimation Kernel filtering algorithms Calculus Robust quaternion adaptive filters quaternion error entropy criterion with fiducial points quaternion errors-in-variables model asymmetric noise

来源：评论

学校读者我要写书评

暂无评论

GSP-KalmanNet: Tracking Graph Signals via Neural-Aided Kalman filtering

引用

IEEE TRANSACTIONS ON SIGNAL PROCESSING 2024年 72卷 3700-3716页

作者： Buchnik, Itay Sagi, Guy Leinwand, Nimrod Loya, Yuval Shlezinger, Nir Routtenberg, Tirza Beng Gurion Univ Negev Sch Elect & Comp Engn ECE IL-84105 Beer Sheva Israel

Dynamic systems of graph signals are encountered in various applications, including social networks, power grids, and transportation. While such systems can often be described as state space (SS) models, tracking graph signals via conventional tools based on the Kalman filter (KF) and its variants is typically challenging. This is due to the nonlinearity, high dimensionality, irregularity of the domain, and complex modeling associated with real-world dynamic systems of graph signals. In this work, we study the tracking of graph signals using a hybrid model-based/data-driven approach. We develop the GSP-KalmanNet, which tracks the hidden graphical states from the graphical measurements by jointly leveraging graph signal processing (GSP) tools and deep learning (DL) techniques. The derivations of the GSP-KalmanNet are based on extending the KF to exploit the inherent graph structure via designing a graph frequency domain filtering and replacing the Kalman gain (KG) with a graph filter that minimizes the prediction error. Thus, it considerably simplifies the computational complexity entailed in processing high-dimensional signals and increases the robustness to small topology changes. Then, we use data to learn the KG, namely, the graph filter, following the recently proposed KalmanNet framework, which copes with partial and approximated modeling, without forcing a specific model over the noise statistics. Restricting the KG to a graph filter in the proposed GSP-KalmanNet reduces learned parameters, thereby enhancing stability. Our empirical results demonstrate that the GSP-KalmanNet achieves enhanced accuracy and run time performance, and improved robustness to model misspecifications compared with both model-based and data-driven benchmarks.

关键词： Complexity theory Computational modeling Kalman filters Frequency-domain analysis filtering algorithms Numerical models Heuristic algorithms Graph signal processing (GSP) graph filters deep learning Kalman filtering

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：