检索结果-内蒙古大学图书馆

Autonomous vehicle surveillance through fuzzy C-means segmentation and DeepSORT on aerial images

PeerJ computer Science 2025年 11卷 1-23页

作者： Qureshi, Asifa Mehmood Alotaibi, Moneerah Alotaibi, Sultan Refa AlHammadi, Dina Abdulaziz Jamal, Muhammad Asif Jalal, Ahmad Lee, Bumshik Department of Computer Science Air University Islamabad Pakistan Department of Computer Science College of Science and Humanities Shaqra University Dawadmi Saudi Arabia Department of Information Systems College of Computer and Information Sciences Princess Nourah bint Abdulrahman University Riyadh Saudi Arabia Information and Communication Engineering Chosun University Gwangju Korea Republic of Department of Computer Science and Engineering College of Informatics Korea University Seoul Korea Republic of

The high mobility of uncrewed aerial vehicles (UAVs) has led to their usage in various computer vision applications, notably in intelligent traffic surveillance, where it enhances productivity and simplifies the process. Yet, there are still several challenges that must be resolved to automate these systems. One significant challenge is the accurate extraction of vehicle foregrounds in complex traffic scenarios. As a result, this article proposes a novel vehicle detection and tracking system for autonomous vehicle surveillance, which employs Fuzzy C-mean clustering to segment the aerial images. After segmentation, we employed the YOLOv4 deep learning algorithm, which is efficient in detecting small-sized objects in vehicle detection. Furthermore, an ID assignment and recovery algorithm based on Speed-Up Robust Feature (SURF) is used for multi-vehicle tracking across image frames. Vehicles are determined by counting in each image to estimate the traffic density at different time intervals. Finally, these vehicles were tracked using DeepSORT, which combines the Kalman filter with deep learning to produce accurate results. Furthermore, to understand the traffic flow direction, the path trajectories of each tracked vehicle is projected. Our proposed model demonstrates a noteworthy vehicle detection and tracking rate during experimental validation, attaining precision scores of 0.82 and 0.80 over UAVDT and KIT-AIS datasets for vehicle detection. For vehicle tracking, the precision is 0.87 over the UAVDT dataset and 0.83 for the KIT-AIS dataset. © 2025 Qureshi et al.

关键词： Image segmentation

来源：评论

学校读者我要写书评

暂无评论

A Two-Pronged Approach to Security of Discrete-Event Systems

引用

Journal of Systems Science & Complexity 2025年第1期38卷 178-209页

作者： MATSUI Shoma RUDIE Karen CAI Kai Department of Electrical and Computer Engineering&the Ingenuity Labs Research Institute Queen's UniversityKingston K7L 3N6Canada Department of Core Informatics Osaka Metropolitan UniversityOsaka 545-0051Japan

This paper investigates a security problem of simultaneously addressing two types of attacks:Eavesdropping and *** authors model the target system as a discrete-event system(DES)with subsets of concealable events and protectable events,in order to make the proposed methodology applicable to various practical systems and employ two existing works of DES security:Degree of opacity and state ***,the authors consider that all protectable events are observable,and some observable events are *** addition,protectable events cannot be protected once they are *** such a constraint,the goal is to figure out which events to conceal and which transitions to protect so that the prescribed requirements of degree of opacity and state protection are *** this work the authors decide which events to conceal as all transitions of a given event label are concealed or not *** proposed problem formulation also requires a solution to only involve absolutely necessary protectable events in order for the system to avoid superfluous protection *** authors first examine a general version of our security problem with an intuitive algorithm to compute acceptable solutions,and then present a special version which results in a reduced computation time compared to the general version.

关键词： Degree of opacity discrete-event systems security state protection

来源：评论

学校读者我要写书评

暂无评论

Deep Learning Quantitative Susceptibility Mapping With Reduced Training Time Based on Local Field Transformation

引用

IEEE Transactions on Emerging Topics in Computational Intelligence 2025年第2期9卷 1527-1538页

作者： Wu, Ming-Long Huang, De-Rong Cheng, Jhih-Shan National Cheng Kung University Department of Computer Science and Information Engineering Tainan701 Taiwan National Cheng Kung University Institute of Medical Informatics Tainan701 Taiwan

A quantitative susceptibility mapping (QSM) approach using single-orientation imaging data is proposed in this study. The proposed method generates local field maps at five predefined orientations via deep learning from single-orientation local field maps, followed by physical dipole inversion using the analytical Calculation of Susceptibility through Multiple Orientation Sampling (COSMOS) algorithm. We termed the method pseudo COSMOS (pCOSMOS). By isolating the physical dipole inversion process from deep learning, training of the neural network model becomes greatly accelerated. Brain imaging data from 10 healthy subjects (7 training and 3 testing) acquired at 3T with five different orientations relative to the main magnetic field were included to evaluate the reconstruction quality. Results compared against two iterative (Morphology Enabled Dipole Inversion or MEDI, and Structural Feature-based Collaborative Reconstruction or SFCR) and two deep learning QSM approaches (QSMnet and Learned Proximal Convolutional Neural Network or LPCNN) using COSMOS as the reference standard revealed that the susceptibility maps obtained from pCOSMOS yielded global quantitative performance comparable to QSMnet and LPCNN, while being prominently superior to MEDI and SFCR. Estimation errors in magnetic susceptibility were between 0.01 and 0.02 ppm voxel-wise, and 0.006 ± 0.004 ppm for five selected gray matter areas. Visual inspection further suggested better preserved tissue boundaries in pCOSMOS than QSMnet and LPCNN. Training time was reduced from 2 days for LPCNN and 12 hours for QSMnet to 3.2 hours for pCOSMOS. It is concluded that the proposed pCOSMOS method is an alternative deep learning QSM approach when computational efficiency is essential. © 2017 IEEE.

关键词： Brain mapping

来源：评论

学校读者我要写书评

暂无评论

Optimizing task scheduling in cloud environments: a hybrid golden search whale optimization algorithm approach

引用

Neural Computing and Applications 2025年 1-23页

作者： Acharya, Biswaranjan Panda, Sucheta Das, Satyabrata Majhi, Santosh Kumar Gerogiannis, Vassilis C. Kanavos, Andreas Department of Computer Science and Engineering Veer Surendra Sai University of Technology Odisha Burla India Bilaspur India Department of Digital Systems University of Thessaly Larissa Greece Department of Informatics Ionian University Corfu Greece

Managing fluctuating workloads and optimizing resource utilization in cloud environments pose significant challenges, particularly in fields requiring real-time data processing, such as healthcare. This paper introduces a novel hybrid metaheuristic technique, the Golden Search Whale Optimization Algorithm (GSWOA), specifically designed for scheduling independent dynamic biomedical data. GSWOA merges the strengths of Golden Search Optimization (GSO) and Whale Optimization Algorithm (WOA), optimizing numerical function optimization and achieving a balance between exploration and exploitation. The algorithm’s effectiveness was assessed using MATLAB by applying standard benchmark functions and further evaluated on a real-world biomedical dataset within the CloudSim environment. The performance evaluations demonstrate that GSWOA significantly outperforms existing metaheuristic and traditional scheduling techniques, achieving a 42.71% increase in resource utilization and a 14.17% reduction in makespan compared to conventional methods. These results highlight GSWOA’s potential to enhance scheduling efficiency substantially in cloud computing infrastructures, suggesting it is a powerful tool for complex task allocations. Future research will explore the scalability of GSWOA and its applicability across other data-intensive sectors. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2025.

关键词： Scheduling algorithms

来源：评论

学校读者我要写书评

暂无评论

(r, k, Ε)-Anonymization: Privacy-Preserving Data Publishing Algorithm Based on Multi-Dimensional Outlier Detection, k-Anonymity, and Ε-Differential Privacy

引用

IEEE Access 2025年 13卷 70422-70435页

作者： Kara, Burak Cem Eyupoglu, Can Karakus, Oktay National Defence University Turkish Air Force Academy Department of Computer Engineering İstanbul34149 Turkey Cardiff University School of Computer Science and Informatics CardiffCF24 4AG United Kingdom

In recent years, there has been a tremendous rise in both the volume and variety of big data, providing enormous potential benefits to businesses that seek to utilize consumer experiences for research or commercial purposes. The general data protection regulation (GDPR) implementation, on the other hand, has introduced extensive control over the use of individuals’ personal information and placed many limits. Data anonymization technologies have become an important solution for businesses trying to generate value from data while adhering to GDPR limitations. To address these challenges, researchers have developed various methods, including k-anonymity and Ε -differential privacy, offering solutions for both industry and academia. However, protecting individuals’ privacy against diverse attack attempts presents significant challenges for anonymization models that rely solely on a single technique, highlighting the need for more adaptable and hybrid approaches. In this study, a new hybrid anonymization algorithm called (r, k, Ε )-anonymization has been proposed, which combines k-anonymity and Ε-differential privacy models in a consistent framework and provides stronger privacy guarantees compared to existing privacy-preserving models. The proposed algorithm is capable of overcoming well-known shortcomings of the k-anonymity and Ε -differential privacy models, and it has been confirmed by extensive tests on real-world datasets. The proposed (r, k, Ε )-anonymization algorithm outperforms k-anonymity and Ε -differential privacy in terms of the average error rate measure, achieving data utility increases of 31.74% and 26.99%, respectively. © 2013 IEEE.

关键词： Anonymity

来源：评论

学校读者我要写书评

暂无评论

Driver Drowsiness Detection Using Swin Transformer and Diffusion Models for Robust Image Denoising

引用

IEEE Access 2025年 13卷 71880-71907页

作者： El-Nabi, Samy Abd Ibrahim, Ahmed F. El-Rabaie, El-Sayed M. Hassan, Osama F. Soliman, Naglaa F. Ramadan, Khalil F. El-Shafai, Walid Faculty of Computer Science and Engineering Department of Artificial Intelligence Engineering South Sinai 46511 Egypt Menoufia University Faculty of Electronic Engineering Department of Electronics and Electrical Communications Engineering Menouf32952 Egypt Suez Canal University Faculty of Computers and Informatics Information Systems Department Ismailia41522 Egypt Princess Nourah bint Abdulrahman University College of Computer and Information Sciences Department of Information Technology P.O. Box 84428 Riyadh11671 Saudi Arabia Computer Science Department Riyadh11586 Saudi Arabia

With the rapid development of intelligent transportation systems and growing emphasis on driver safety, real-time detection of driver drowsiness has become a critical area of research. This study presents a robust and scalable driver drowsiness detection framework that integrates a Swin Transformer-based deep learning model with a diffusion model for image denoising. While conventional convolutional neural networks (CNNs) are effective in standard vision tasks, they often suffer performance degradation in real-world driving scenarios due to noise, poor lighting, motion blur, and adversarial attacks. To address these challenges, the proposed model focuses on eye-state detection, specifically, prolonged eye closure, as a primary indicator of driver disengagement and fatigue. Our system introduces a novel preprocessing stage using a denoising diffusion model built on a U-Net encoder-decoder architecture, effectively mitigating the impact of Gaussian noise and adversarial perturbations. Additionally, we incorporate adversarial training with Fast Gradient Sign Method (FGSM) and Projected Gradient Descent (PGD) attacks, demonstrating significant improvements in classification accuracy and resilience. Evaluations are conducted on two benchmark datasets, Eye-Blink and Closed Eyes in the Wild (CEW), under both clean and noisy conditions. Comparative experiments show that the proposed system outperforms several state-of-the-art models, including ViT, ResNet50V2, InceptionV3, MobileNet, DenseNet169, and VGG19, in terms of accuracy (up to 99.82%), PSNR (up to 41.61 dB), and SSIM (up to 0.984), while maintaining competitive inference times suitable for practical deployment. Moreover, a detailed sensitivity analysis of data augmentation strategies reveals that techniques such as rotation and horizontal flip substantially enhance the model’s generalization across variable visual inputs. The system also demonstrates improved robustness under real-world black-box scenarios and adver

关键词： Sensitivity analysis

来源：评论

学校读者我要写书评

暂无评论

Optimal control for unreliable manufacturing and remanufacturing system under environmental subsidy and overrun penalty mechanism

引用

IEEE Transactions on engineering Management 2025年 72卷 1983-2002页

作者： Turki, Sadok Jemmali, Mahdi Hidri, Lotfi University of Lorraine Department of Logistic and Maintenance UFR MIM at Metz Metz17 France University of Sharjah Computer Science Department College of Computing and Informatics Sharjah United Arab Emirates King Saud University Industrial Engineering Department College of Engineering Riyadh11421 Saudi Arabia

Because of environmental concerns, remanufacturing has become an integral process for many production companies. Most published papers dealing with manufacturing and remanufacturing systems (MRSs) overlook some industrial features. These include machine unreliability, the distinction between newly manufactured and remanufactured products, and uncertainty about the returned used products (RUPs). As a result, these models do not reflect the reality of industrial systems. This study aims to bridge the gap in research on MRSs by elaborating an optimal and dynamic control of manufacturing, remanufacturing, and recovery of RUPs. The goal is to develop a model based on manufacturing remanufacturing emissions hedging point control policy (MREHP) to adjust the production of new and remanufactured products, and recovery of RUPs according to economic and environmental thresholds. To reduce carbon emissions, a new method is proposed based on the environmental subsidy and overrun penalty mechanism (ESOPM), and that is combined with MREHP to control in parallel the MRS and generated emissions. A robust optimization algorithm is developed to determine the optimal economic and environmental thresholds in MREHP and ESOPM. A comparative study is provided to evaluate our proposed ESOPM in comparison with traditional reduction policies: carbon cap and trade (CCT) and carbon tax regulation (CTR). The obtained results reveal that machines' reliability, rate of RUPs, emissions subsidy, and carbon emissions limit have notable impacts on economic and environmental thresholds. Additionally, results show that the proposed ESOPM performs higher profit than CTR and CCT while reducing emissions when the carbon price is below a certain value. © 1988-2012 IEEE.

关键词： Fabrication

来源：评论

学校读者我要写书评

暂无评论

LSTM-XGBoost: An Ensemble Model for Blood Demand Distribution Forecasting – A Case Study in Zakho City, Kurdistan Region, Iraq

引用

Operations Research Forum 2025年第1期6卷 1-22页

作者： Zebari, Rizgar R. Zebari, Gheyath M. Al-zebari, Adel Mohammed, Marwan Aziz Department of Computer Science College of Science Knowledge University Erbil 44001 Iraq Department of Information Technology Technical College of Informatics Akre University for Applied Sciences Kurdistan Akre Iraq Department of Computer Engineering College of Engineering Knowledge University Erbil 44001 Iraq

A safe and adequate blood supply is essential for healthcare systems to function effectively. Accurately forecasting blood demand plays a key role in efficient inventory management and resource allocation. Traditional forecasting methods, like moving averages and ARIMA models, often fall short due to the complexity of factors influencing blood demand. This study introduces an innovative hybrid ensemble model that combines Long Short-Term Memory (LSTM) networks with XGBoost, harnessing their combined strengths to enhance forecasting accuracy. By analyzing blood donation data from the Zakho Blood Bank (January 1, 2015—July 22, 2022), the model outperforms individual LSTM and XGBoost models, excelling in metrics such as Mean Square Error (MSE) and Mean Absolute Error (MAE). These findings underscore the potential of advanced machine learning techniques to improve healthcare supply chain management and ensure the timely availability of critical blood supplies. © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2025.

关键词： Blood banking Blood demand forecasting Healthcare supply chain LSTM network Machine learning XGBoost

来源：评论

学校读者我要写书评

暂无评论

Method of Automated Dataset Collection for Microwave Filters Synthesis

Method of Automated Dataset Collection for Microwave Filters...

引用

2025 Systems of Signals Generating and Processing in the Field of on Board Communications, SOSG 2025

作者： Arinin, O.V. Bakhmach, D.M. Katsnelson, A.I. Kurbeteva, A.S. Yelizarov, A.A. Moscow Technical University of Communications and Informatics Moscow Russia Hse University Electronic Engineering Department Moscow Russia Hse University Computer Engineering Department Moscow Russia

ISBN: (纸本)9798331532635

This research discusses the method of dataset collection automatization for microwave filter synthesis by integrating machine learning techniques, thus reducing development time. Utilizing the 3D electromagnetic analysis software package, the study involves simulation and collecting geometric parameters and amplitude-frequency characteristics from three variants of passband highly selective microstrip tworesonator combined filters with stepped impedance resonators. In this paper, specific items and restrictions on filter topology creation in the simulation program are discussed, as well the review of the script used for dataset collection and its requirements is given. In addition, a dataset structure and format are described. © 2025 IEEE.

关键词： Software packages

来源：评论

学校读者我要写书评

暂无评论

Face image authentication scheme based on MTCNN and SLT

引用

Multimedia Tools and Applications 2025年 1-43页

作者： Thabit, Rasha Al-Askari, Mohanad A. Mohammed, Dunya Zeki Anaam, Elham Abdulwahab Mahmood, Zainab H. Jabbar, Dina Jamal Salih, Zahraa Aqeel Information System Department College of Computer Sciences and Information Technology University of Al-Anbar Al-Anbar Iraq Computer Engineering Department College of Engineering Al-Iraqia University Baghdad Iraq Electronic and Communications Engineering Department College of Engineering Gilgamesh University Baghdad Iraq Faculty of Computing and Informatics Multimedia University Selangor Cyberjaya Malaysia Department of Cybersecurity Engineering College of Information Engineering Al-Nahrain University Baghdad Iraq Electrical Engineering Technical Collage Middle Technical University Baghdad Iraq Department of Business Administration Alamaal University College Baghdad Iraq

DeepFakes and face image manipulation methods have been widely distributed in the last few years and several techniques have been presented to check the authenticity of the face image and detect manipulation if exists. Most of the available manipulation detection techniques have been successfully applied to reveal one type of manipulation under specific conditions, however, many limitations and challenges can be encountered in this field. To overcome some limitations and challenges, this paper presents a new face image authentication (FIA) scheme based on Multi-Task Cascaded Conventional Neural Networks (MTCNN) and watermarking in Slantlet transform (SLT) domain. The proposed FIA scheme has three main algorithms that are face detection and selection, embedding, and extraction algorithms. Different block sizes have been used to divide the image into non-overlapping blocks followed by classifying them into two groups that are blocks from face area (FA) and blocks from the remaining area (RA) of the image. In the embedding algorithms, the authentication information is generated from FA blocks and embedded in the RA blocks. In the extraction algorithms, the embedded information is extracted from RA blocks and compared with the calculated data from FA blocks to reveal manipulations and localize the manipulated blocks if exist. Extensive experiments have been conducted to evaluate the performance of the proposed FIA scheme for different face images. The experimental work included tests for payload, capacity, visual quality, time complexity, and localization of manipulations. The results proved the efficiency of the proposed scheme in detecting and localizing different face image manipulations such as attributes attacks, retouching attacks, expression swap, face swap, and morphing attacks. The proposed scheme overcomes many limitations and it is 100% accurate in localizing the tampered blocks which makes it a better candidate for practical applications. © The Author(s), un

关键词： Authentication

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

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

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

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

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：