检索结果-内蒙古大学图书馆

Enhancing Object Detection With Fourier Series

IEEE TRANSACTIONS ON pattern analysis AND MACHINE INTELLIGENCE 2025年第4期47卷 2581-2596页

作者： Liu, Jin Lu, Zhongyuan Cen, Yaorong Hu, Hui Shao, Zhenfeng Hong, Yong Jiang, Ming Xu, Miaozhong Wuhan Univ State Key Lab Informat Engn Surveying Mapping & Re Wuhan 430079 Peoples R China Wuhan Xiongchu Gaojing Technol Co Ltd Wuhan 430000 Peoples R China Wuhan Univ Elect Informat Sch Wuhan 430079 Peoples R China Aviat Ind Corp China Res Inst Leihua Elect Technol Beijing 100028 Peoples R China

Traditional object detection models often lose the detailed outline information of the object. To address this problem, we propose the Fourier Series Object Detection (FSD). It encodes the object's outline closed curve into two one-dimensional periodic Fourier series. The Fourier Series Model (FSM) is constructed to regress the Fourier series for each object in the image. Thus, during inference, the detailed outline information of each object can be retrieved. We introduce Rolling Optimization Matching for Fourier loss to ensure that the model's learning process is not affected by the sequence of the starting points of the labeled contour points, speeding up the training process. The FSM demonstrates improved feature extraction and descriptive capabilities for non-rectangular or elongated object regions. The model achieves AP50 = 73.3% on the DOTA 1.5 dataset, which surpasses the state-of-the-art (SOTA) method by 6.44% at 66.86%. On the UCAS dataset, the model achieves AP50 = 97.25%, also surpassing the performance indicators of the SOTA methods. Furthermore, we introduce the object's Fourier power spectrum to describe outline features and the Fourier vector to indicate its direction. This enhances the scene semantic representation of the object detection model and paves a new pathway for the evolution of object detection methodologies.

关键词： Fourier series Object detection Shape mathematical models Transformers Optimization Vectors Predictive models image segmentation Gaussian distribution closed curve of arbitrary shape and directional vector

来源：评论

学校读者我要写书评

暂无评论

Speed and Position Measurement of Rotating Machinery With a Triangular pattern Vision Encoder

引用

IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT 2025年 74卷

作者： Thielemans, Yentl Kayedpour, Nezmin Coene, Annelies Crevecoeur, Guillaume De Kooning, Jeroen D. M. Dynam Syst & Control Grp DySC B-8500 Kortrijk Belgium Univ Ghent Dept Electromech Syst & Met Engn Energy & Syst Lab EnSy Tech Lane Ghent Sci Pk-Campus A B-9052 Ghent Belgium FlandersMakeUGent Corelab MIRO Flanders Make B-3001 Heverlee Belgium

The measurement of the position and speed of rotating machinery is essential in the manufacturing industry in order to effectively control a process. Traditional measurement methods such as encoders, resolvers, and tachometers are limited to a single rotating element, require precise (mechanical) installation, and measure on the motor shaft rather than the load. This article proposes a novel methodology to overcome the shortcomings of conventional sensors with a triangular pattern vision encoder. The approach uses single pixellines in combination with a triangular pattern applied to a roller to capture the position and speed of one or more axes simultaneously. First, a preprocessing step detects the position of a roller in the frame. Second, the width of the triangle is detected and mapped to accurately measure the position of the roller. The main contributions are improved robustness compared to state-of-theart techniques, straightforward multitarget velocity analysis, and the integration of FDZP as a subpixel technique. The proposed design is validated on an industrial web processing machine (WPM) and achieves an accuracy of 480 trad, an improvement of approximately 38% over the benchmark incremental optical encoder, which achieves an accuracy of 770 trad.

关键词： image edge detection Accuracy Rotation measurement Position measurement mathematical models Cameras Shafts Monitoring Tachometers Standards rotating machinery sensing and instrumentation speed measurement vision-based encoder

来源：评论

学校读者我要写书评

暂无评论

Transforming tabular data into images via enhanced spatial relationships for CNN processing

引用

SCIENTIFIC REPORTS 2025年第1期15卷 1-14页

作者： Alenizy, Hameedah A. Berri, Jawad King Saud Univ Coll Comp & Informat Sci Informat Syst Dept Riyadh 11451 Saudi Arabia Princess Nourah bint Abdulrahman Univ Appl Coll Dept Comp Sci programs Riyadh 13414 Saudi Arabia

Convolutional neural networks (CNNs), renowned for their efficiency in image analysis, have revolutionized pattern and structure recognition in visual data. Despite their success in image-based applications, CNNs face challenges when applied to tabular data due to the lack of inherent spatial relationships among features. This weakness can be overcome if the original tabular data is expanded to create an enhanced image that exhibits pseudo-spatial relationships. This paper introduces an original approach that transforms tabular data into a format suitable for CNN processing. The Novel Algorithm for Convolving Tabular Data (NCTD) applies mathematical transformations including rotation translation and reflection, to simulate spatial relationships within the data, thereby constructing a data structure analogous to a 2D synthetic image. This transformation enables CNNs to process tabular data efficiently by leveraging automated feature extraction and enhanced pattern recognition. The NCTD algorithm was extensively evaluated and compared with traditional machine learning algorithms and existing methods on ten benchmark datasets. The results showed that NCTD consistently surpassed the majority of competing algorithms in nine out of ten datasets, indicating its potential as a robust tool for extending CNN applicability beyond conventional image-based domains, particularly in complex classification and prediction.

关键词： NCTD Tabular data image generation Convolutional neural networks image classification

来源：评论

学校读者我要写书评

暂无评论

Range-Null Space Decomposition With Frequency-Oriented Mamba for Spectral Superresolution

引用

IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING 2025年 18卷 10292-10306页

作者： Weng, Meimei Liu, Jianjun Yang, Jinlong Wu, Zebin Xiao, Liang Jiangnan Univ Sch Artificial Intelligence & Comp Sci Jiangsu Prov Engn Lab Pattern Recognit & Computat Wuxi 214122 Peoples R China Nanjing Univ Sci & Technol Sch Comp Sci Nanjing 210094 Peoples R China

Spectral superresolution (SSR) is a technique aimed at reconstructing hyperspectral images (HSIs) from images with low spectral resolution. Previous methods combining mathematical models with deep learning have shown promising performance for HSI reconstruction. However, these methods still have limitations when dealing with complex scenes, especially in terms of data consistency and realness. To address these issues, we propose a model-driven SSR network that integrates range-null space decomposition with deep learning. Specifically, we solve for the range space (R-Space) part and null space (N-Space) part to reconstruct the desired HSI with consistency and realness. The R-Space is primarily iteratively derived from the input multispectral image to ensure reliable data consistency, while the N-Space reflects the true distribution of the target HSI, and its proper representation helps improve visual quality. To enhance N-Space exploration, we construct a frequency-oriented N-Space learning module that leverages Mamba and self-attention to separately extract spatial and spectral information in the frequency domain. In addition, we introduce a structure tensor term and a multikernel maximum mean discrepancy term in the loss function to constrain R-Space and N-space, respectively. Experimental results show that the proposed method achieves excellent performance.

关键词： image reconstruction Transformers Superresolution Hyperspectral imaging Feature extraction Frequency-domain analysis Data mining Convolution Visualization Optimization Hyperspectral image (HSI) Mamba range-null space decomposition (RNSD) spectral superresolution (SSR)

来源：评论

学校读者我要写书评

暂无评论

Unsupervised Feature Matching for Affine Histological image Registration 27th

Unsupervised Feature Matching for Affine Histological Image...

引用

27th International Conference on pattern Recognition, ICPR 2024

作者： Pyatov, Vladislav A. Sorokin, Dmitry V. Laboratory of Mathematical Methods of Image Processing Faculty of Computational Mathematics and Cybernetics Lomonosov Moscow State University Moscow Russia

ISBN: (纸本)9783031782008

One of the most common tasks in histopathology is the visual comparison of the images of successive multiply stained tissue sections. Automatic image registration is crucial to perform this analysis. Although the tissue sections in general undergo non-rigid deformations, the initial linear image alignment impacts the overall registration drastically. However, most of the recent works do not study the linear transformation compensation separately and focus on the non-linear part. In this work, we propose a novel unsupervised feature matching approach for affine registration of histological images. We perform the evaluation on the Automatic Non-rigid Histological image Registration (ANHIR) dataset and show the supremacy of our method over the existing affine registration approaches in therms of accuracy and robustness. The code is available at https://***/VladPyatov/UnFeMa. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： image registration

来源：评论

学校读者我要写书评

暂无评论

analysis of mechanical behavior of fiber-glass plastic with hole pattern using digital image correlation and acoustic emission methods

引用

FRATTURA ED INTEGRITA STRUTTURALE-FRACTURE AND STRUCTURAL INTEGRITY 2024年第68期18卷 63-76页

作者： Strungar, E. M. Lobanov, D. S. Chebotareva, E. A. Kochneva, Y., V Perm Natl Res Polytech Univ Ctr Expt Mech Perm Russia

In this paper, tensile tests of specimens with a pattern of holes made of fiber-glass plastic based on combined epoxy and phenol-formaldehyde resins are carried out in order to study the processes of damage accumulation and tension fracture. The Vic-3D video system is used to evaluate damage development and inhomogeneity of strain localization during loading. Continuous recording of acoustic emission signals is carried out during the tests, resulting in obtaining data on fracture mechanisms in the material. Ranges of peak frequencies are identified. Surface analysis of specimens was carried out using a microscope. A significant reduction in strength occurs due to the presence of a circular hole in the material, although additional holes do not exacerbate this effect. Fracture patterns of specimens with a hole pattern have been analyzed, and different "paths" of fracture have been observed. The comparison of strain fields obtained on the basis of application of three-dimensional digital optical system with the configuration of strain fields constructed as a result of numerical modeling by the finite element method has been carried out. It is found that the strain fields for different open hole patterns are quantitatively and qualitatively similar and identical. © 2024 The Author(s).

关键词： Digital image Correlation method Acoustic emission mathematical modeling Stress concentrator Open-hole Fiberglass

来源：评论

学校读者我要写书评

暂无评论

A comprehensive survey of golden jacal optimization and its applications

引用

COMPUTER SCIENCE REVIEW 2025年 56卷

作者： Hosseinzadeh, Mehdi Tanveer, Jawad Rahmani, Amir Masoud Alanazi, Abed Zaidi, Monji Mohamed Aurangzeb, Khursheed Alinejad-Rokny, Hamid Porntaveetus, Thantrira Lee, Sang-Woong Duy Tan Univ Sch Comp Sci Da Nang Vietnam Jadara Univ Jadara Univ Res Ctr Irbid Jordan Sejong Univ Dept Comp Sci & Engn Seoul 05006 South Korea Natl Yunlin Univ Sci & Technol Future Technol Res Ctr Yunlin Taiwan Prince Sattam Bin Abdulaziz Univ Coll Comp Engn & Sci Dept Comp Sci POB 151 Al Kharj 11942 Saudi Arabia King Khalid Univ Coll Engn Dept Elect Engn Abha 61421 Saudi Arabia King Khalid Univ Ctr Engn & Technol Innovat Abha 61421 Saudi Arabia King Saud Univ Coll Comp & Informat Sci Dept Comp Engn POB 51178 Riyadh 11543 Saudi Arabia UNSW Sydney Grad Sch Biomed Engn UNSW BioMed Machine Learning Lab BML Sydney NSW 2052 Australia Chulalongkorn Univ Ctr Excellence Genom & Precis Dent Geriatr Dent & Special Patients Care Int Program Clin Res CtrDept PhysiolFac Dent Bangkok Thailand Gachon Univ Pattern Recognit & Machine Learning Lab Seongnam 13120 South Korea

In recent decades, there has been an increasing interest from the research community in various scientific and engineering fields, including robotic control, signal processing, image processing, feature selection, classification, clustering, and other issues. Many optimization problems are inherently complicated and complex. They cannot be solved by traditional optimization methods, such as mathematical programming, because most conventional optimization methods focus on evaluating first derivatives. On the other hand, metaheuristic algorithms have high ability and adaptability in finding near-optimal solutions in a reasonable time for different optimization problems due to parallel search and balance between exploration and exploitation. This study discusses the basic principles and mechanisms of the GJO algorithm and its challenges. This review aims to provide valuable insights into the potential of the GJO algorithm for real-world and scientific optimization tasks. In this paper, a complete review of the Golden Jackal Optimization (GJO) algorithm for various optimization problems is done. The GJO algorithm is one of the metaheuristic algorithms invented in 2022 and inspired by the life of natural jackals. This paper's complete classification of GJO in hybrid, improved, binary, multi-objective, and optimization problems is done. The analysis shows that the percentage of studies conducted in the four fields of hybrid, improved variants of GJO (binary, multi-objective), and optimization are 11 %, 44 %, 9 %, and 36 %, respectively. Studies have shown that this algorithm performs well in real-world challenges. GJO is a powerful tool for solving scientific and engineering problems flexibly.

关键词： Optimization problems Metaheuristic algorithms Golden Jackal Optimization Improved

来源：评论

学校读者我要写书评

暂无评论

Adversarial Attack on YOLO Neural Network

Adversarial Attack on YOLO Neural Network

引用

International Russian Smart Industry Conference (SmartIndustryCon)

作者： Nikolai V. Teterev Vladislav E. Trifonov Alla B. Levina Saint Petersburg Electrotechnical University “LETI” Saint Petersburg Russian Federation

ISBN: (数字)9798331511241

ISBN: (纸本)9798331511258

This paper describes and demonstrates a comprehensive analysis of structured criteria of formalized conditions for creating universal images falsely classified by computer vision algorithms called adversarial examples based on YOLO neural network models. In this paper, a pattern was identified and studied using the above mathematical model of the proposed algorithm for the successful creation of a universal destructive image depending on the generated dataset, on which neural networks were trained using a fast sign gradient attack. This pattern is demonstrated for YOLO 8, YOLO 9, YOLO 10, and YOLO 11 classifier models trained on the basis of the standard COCO dataset.

关键词： YOLO Industries Computer vision Gradient methods Computational modeling Neural networks mathematical models Classification algorithms Internet Standards

来源：评论

学校读者我要写书评

暂无评论

From Missing Pieces to Masterpieces: image Completion with Context-Adaptive Diffusion

引用

IEEE Transactions on pattern analysis and Machine Intelligence 2025年 PP卷 PP页

作者： Shamsolmoali, Pourya Zareapoor, Masoumeh Zhou, Huiyu Felsberg, Michael Tao, Dacheng Li, Xuelong University of York Department of Computer Science United Kingdom University of Leicester School of Computing and Mathematical Sciences United Kingdom Linkoping University Computer Vision Laboratory Sweden Nanyang Technological University College of Computing & Data Science Singapore China Northwestern Polytechnical University Key Laboratory of Intelligent Interaction and Applications Ministry of Industry and Information Technology Xi'an China

image completion is a challenging task, particularly when ensuring that generated content seamlessly integrates with existing parts of an image. While recent diffusion models have shown promise, they often struggle with maintaining coherence between known and unknown (missing) regions. This issue arises from the lack of explicit spatial and semantic alignment during the diffusion process, resulting in content that does not smoothly integrate with the original image. Additionally, diffusion models typically rely on global learned distributions rather than localized features, leading to inconsistencies between the generated and existing image parts. In this work, we propose ConFill, a novel framework that introduces a Context-Adaptive Discrepancy (CAD) model to ensure that intermediate distributions of known and unknown regions are closely aligned throughout the diffusion process. By incorporating CAD, our model progressively reduces discrepancies between generated and original images at each diffusion step, leading to contextually aligned completion. Moreover, ConFill uses a new Dynamic Sampling mechanism that adaptively increases the sampling rate in regions with high reconstruction complexity. This approach enables precise adjustments, enhancing detail and integration in restored areas. Extensive experiments demonstrate that ConFill outperforms current methods, setting a new benchmark in image completion. © 1979-2012 IEEE.

关键词： Computer aided design

来源：评论

学校读者我要写书评

暂无评论

Quaternion CNN With Salient Features for Color image Denoising

Quaternion CNN With Salient Features for Color Image Denoisi...

引用

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Yi Liu Qiyu Jin Jie Yang School of Mathematical Science Inner Mongolia University Hohhot China Institute of Image Processing and Pattern Recognition Shanghai Jiao Tong University Shanghai China

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

Deep convolutional neural networks have significantly advanced color image denoising. However, existing models often apply grayscale denoising techniques to color images without accounting for inter-channel correlations, resulting in color distortion, detail loss, and visual artifacts. Moreover, these models frequently neglect salient features within convolutional maps. To address these issues, we propose a quaternion CNN model that captures channel correlations and extracts salient features, thereby enhancing color image denoising performance. Specifically, we convert color images into quaternion matrices to better capture these correlations and design a quaternion convolutional network to learn relevant features. Furthermore, an aggregated feature block is introduced to enhance the extraction of salient features and further refine the denoising process. Experimental results on multiple datasets demonstrate that the proposed model achieves superior performance compared to recent state-of-the-art methods.

关键词： Visualization Correlation Acoustic distortion image color analysis Convolution Quaternions Noise reduction Color Feature extraction Convolutional neural networks

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：