检索结果-内蒙古大学图书馆

Single-image super-resolution via a lightweight convolutional neural network with improved shuffle learning

signal image AND VIDEO processing 2024年第1期18卷 233-241页

作者： Lu, Xinbiao Xie, Xupeng Ye, Chunlin Xing, Hao Liu, Zecheng Chen, Yudan Hohai Univ Sch Energy & Elect Engn Nanjing 211100 Peoples R China

With the development of deep learning techniques, single-image super-resolution methods based on deep learning have made great progress, enabling significant improvements in image quality and detail reproduction. However, deep convolutional neural networks are often complicated and hard to be understood, and the computational cost limits the application of the models in practical situations. In order to deploy the network on mobile devices with very limited computing power, we build a refined image super-resolution model based on shuffle learning. Based on extensive experimental results on image super-resolution using three widely used datasets, our model not only achieves high scores on the peak signal-to-noise ratio/structural similarity index matrix, but also is simpler and easier to be implemented than other image super-resolution models.

关键词： Super-resolution Convolutional neural network Model lightweight Shuffle units

来源：评论

学校读者我要写书评

暂无评论

Quantum edge detection of medical images using novel enhanced quantum representation and hill entropy approach

引用

signal image AND VIDEO processing 2024年第2期18卷 1803-1819页

作者： Chaduvula, Kavitha Indira, D. N. V. S. L. S. Markapudi, Baburao Kalyanapu, Srinivas Seshadri Rao Gudlavalleru Engn Coll Dept Informat Technol Gudlavalleru 521356 Andhra Pradesh India Seshadri Rao Gudlavalleru Engn Coll Dept Comp Sci & Engn Gudlavalleru 521356 Andhra Pradesh India Seshadri Rao Gudlavalleru Engn Coll Dept Artificial Intelligence & Data Sci Gudlavalleru 521356 Andhra Pradesh India

Cutting-edge medical image analysis, driven by quantum-based techniques, offers automated information extraction from images, revolutionizing health care. Traditional methods are being outpaced by the demand for advanced real-time digital image processing. This article introduces an innovative approach to medical image edge detection based on entropy. In recent years, various quantum representation models have emerged, addressing the complex nature of medical images characterized by dark backgrounds and low contrast. To enhance image quality, the article introduces the novel enhanced quantum representation model, which leverages the colour operations of Caraiman's quantum image representation model to improve the greyscale values of individual pixels. However, the article acknowledges that quantum noise remains a challenge in image processing due to statistical fluctuations in medical imaging. To combat this, the article introduces a neural network-based hybrid filter, comprising neural edge enhancers and bilateral filters. The neural filter acts as a fusion operator, effectively eliminating quantum noise from the output image. Another challenge addressed in this work is the time complexity of edge detection. The article presents a novel methodology for edge extraction based on Hill entropy for medical images, which involves segmenting the image into objects and backgrounds using a threshold value. This method aims to reduce computation time while producing high-quality edge detection. The proposed algorithm is implemented using MATLAB software and evaluated on various images. The results demonstrate the algorithm's effectiveness, with a notably higher peak signal-to-noise ratio of 41.5312%, a lower mean square error of 0.0214%, and an improved contrast-to-noise ratio of 42.59%. These outcomes underscore the algorithm's superior performance in edge detection for medical images, offering a remarkable accuracy of 97.5% compared to traditional methods.

关键词： Quantum image processing Edge extraction Quantum representation model neural network Hill entropy Hybrid filter Quantum noise

来源：评论

学校读者我要写书评

暂无评论

Survey of new applications of geometric algebra

引用

MATHEMATICAL methods IN THE APPLIED SCIENCES 2024年第14期47卷 11368-11384页

作者： Hitzer, Eckhard Kamarianakis, Manos Papagiannakis, George Vasik, Petr Int Christian Univ Coll Liberal Arts Tokyo Japan Univ Crete Dept Comp Sci Iraklion Greece Brno Univ Technol Fac Mech Engn Brno Czech Republic Int Christian Univ Osawa 3-10-2 Mitaka 1818585 Japan

This survey introduces 101 new publications on applications of Clifford's geometric algebras (GAs) newly published during 2022 (until mid-January 2023). The selection of papers is based on a comprehensive search with ***, followed by detailed screening and clustering. Readers will learn about the use of GA for mathematics, computation, surface representations, geometry, image, and signal processing, computing and software, quantum computing, data processing, neural networks, medical science, physics, electric engineering, control and robotics.

关键词： Clifford geometric algebra computing control data processing electric engineering geometry image and signal processing mathematics medical science neural networks physics quantum computing robotics software surface representations

来源：评论

学校读者我要写书评

暂无评论

Learning stochastic Graph neural Networks With Constrained Variance

引用

IEEE TRANSACTIONS ON signal processing 2023年 71卷 358-371页

作者： Gao, Zhan Isufi, Elvin Univ Cambridge Dept Comp Sci & Technol Cambridge CB3 0FD England Delft Univ Technol Dept Intelligent Syst NL-2628 Delft Netherlands

stochastic graph neural networks (SGNNs) are information processing architectures that learn representations from data over random graphs. SGNNs are trained with respect to the expected performance, which comes with no guarantee about deviations of particular output realizations around the optimal expectation. To overcome this issue, we propose a variance-constrained optimization problem for SGNNs, balancing the expected performance and the stochastic deviation. An alternating primal-dual learning procedure is undertaken that solves the problem by updating the SGNN parameters with gradient descent and the dual variable with gradient ascent. To characterize the explicit effect of the variance-constrained learning, we analyze theoretically the variance of the SGNN output and identify a trade-off between the stochastic robustness and the discrimination power. We further analyze the duality gap of the variance-constrained optimization problem and the converging behavior of the primal-dual learning procedure. The former indicates the optimality loss induced by the dual transformation and the latter characterizes the limiting error of the iterative algorithm, both of which guarantee the performance of the variance-constrained learning. Through numerical simulations, we corroborate our theoretical findings and observe a strong expected performance with a controllable variance.

关键词： stochastic processes Perturbation methods Graph neural networks Training Topology Stability criteria Recommender systems stochastic graph neural networks variance constraint primal-dual learning duality gap convergence

来源：评论

学校读者我要写书评

暂无评论

Manifold and Sparse Low-Rank Unfolding Network for image Compressed Sensing

引用

IEEE signal processing LETTERS 2025年 32卷 1196-1200页

作者： Lin, Leping Zhang, Zhengtai Ouyang, Ning Guilin Univ Elect Technol Sch Informat & Commun Guilin 541004 Peoples R China

Low-rank matrix approximation has been widely applied in image compressed sensing. However, existing methods often suffer from high computational cost and the need for manual parameter tuning. In this letter, we propose a manifold and sparse low-rank unfolding network (MSLR-Net). We design a pyramid rank residual tensor learning module that approximates low-rank matrices by summing the weights of several rank-1 component tensors. Additionally, we introduce a sparse module based on convolutional dictionary learning, which explicitly enforces sparse priors and explores non-local topological low-rank features in a low-dimensional manifold space using graph convolution. The proposed MSLR-Net integrates optimization methods into an iterative unfolding framework within a neural network, offering an end-to-end structure with high interpretability and easy extensibility to other applications. Extensive experiments show that MSLR-Net demonstrates the effectiveness of its three priors and outperforms state-of-the-art approaches.

关键词： Tensors Convolution Manifolds image reconstruction Training Dictionaries Tuning Sparse matrices neural networks Vectors Compressed sensing image reconstruction manifold learning low-rank prior

来源：评论

学校读者我要写书评

暂无评论

A deep neural network for hand gesture recognition from RGB image in complex background

引用

signal image AND VIDEO processing 2024年第SUPPL 1期18卷 861-872页

作者： Tsai, Tsung-Han Ho, Yuan-Chen Chi, Po-Ting Chen, Ting-Jia Natl Cent Univ Dept Elect Engn Taoyuan Taiwan

Deep learning research has gained significant popularity recently, finding applications in various domains such as image preprocessing, segmentation, object recognition, and semantic analysis. Deep learning has gradually replaced traditional algorithms such as color-based methods, contour-based methods, and motion-based methods. In the context of hand gesture recognition, traditional algorithms heavily rely on depth information for accuracy, but their performance is often subpar. This paper introduces a novel approach using a deep neural network for hand gesture recognition, requiring only a single complementary metal oxide semiconductor (CMOS) camera to operate amidst complex backgrounds. The neural network design incorporates depthwise separable convolutional layers, dividing the model into segmentation and recognition components. As our proposed single-stage model, we avoid the use of the whole model and thus reduce the number of weights and calculations. Additionally, in the training phase, the data augmentation and iterative training strategy further increase recognition accuracy. The results show that the proposed work uses little parameter usage while still having a higher gesture recognition rate than the other works.

关键词： Hand gesture recognition Hand segmentation Deep neural network Attention model Depthwise separable convolution Human-computer interaction

来源：评论

学校读者我要写书评

暂无评论

Aircraft Position Estimation Using Deep Convolutional neural Networks for Low SNR (signal-to-Noise Ratio) Values

引用

SENSORS 2025年第1期25卷 97-97页

作者： Mazurek, Przemyslaw Chlewicki, Wojciech West Pomeranian Univ Technol Szczecin Dept Signal Proc & Multimedia Engn al Piastow 17 PL-70310 Szczecin Poland

The safety of the airspace could be improved by the use of visual methods for the detection and tracking of aircraft. However, in the case of the small angular size of airplanes and the high noise level in the image, sufficient use of such methods might be difficult. By using the ConvNN (Convolutional neural Network), it is possible to obtain a detector that performs the segmentation task for aircraft images that are very small and lost in the background noise. In the learning process, a database of actual aircraft images was used. Using the Monte Carlo method, four types of Max algorithms, i.e., Pixel Value, Min. Pixel Value, and Max. Abs. Pixel Value, were compared with ConvNN's forward architecture. The obtained results showed superior detection with ConvNN. For example, if the standard deviation equals 0.1, it was twice as large. Deep dream analysis for network layers is presented, which shows a preference for images with horizontal contrast lines. The proposed solution uses the processed image values for the tracking process with the raw data using the Track-Before-Detect method.

关键词： machine vision air surveillance deep convolutional neural networks image processing

来源：评论

学校读者我要写书评

暂无评论

Hardware-Independent Deep signal processing: A Feasibility Study in Echocardiography

引用

IEEE TRANSACTIONS ON ULTRASONICS FERROELECTRICS AND FREQUENCY CONTROL 2024年第11期71卷 1491-1500页

作者： Gundersen, Erlend Loland Smistad, Erik Jahren, Tollef Struksnes Masoy, Svein-Erik GE Vingmed Ultrasound AS N-3183 Horten Norway Norwegian Univ Sci & Technol Dept Circulat & Med Imaging N-7491 Trondheim Norway SINTEF Hlth N-7052 Trondheim Norway

Deep learning (DL) models have emerged as alternative methods to conventional ultrasound (US) signal processing, offering the potential to mimic signal processing chains, reduce inference time, and enable the portability of processing chains across hardware. This article proposes a DL model that replicates the fine-tuned BMode signal processing chain of a high-end US system and explores the potential of using it with a different probe and a lower end system. A deep neural network (DNN) was trained in a supervised manner to map raw beamformed in-phase and quadrature component data into processed images. The dataset consisted of 30 000 cardiac image frames acquired using the GE HealthCare Vivid E95 system with the 4Vc-D matrix array probe. The signal processing chain includes depth-dependent bandpass filtering, elevation compounding, frequency compounding, and image compression and filtering. The results indicate that a lightweight DL model can accurately replicate the signal processing chain of a commercial scanner for a given application. Evaluation on a 15-patient test dataset of about 3000 image frames gave a structural similarity index measure (SSIM) of 98.56 +/- 0.49. Applying the DL model to data from another probe showed equivalent or improved image quality. This indicates that a single DL model may be used for a set of probes on a given system that targets the same application, which could be a cost-effective tuning and implementation strategy for vendors. Furthermore, the DL model enhanced image quality on a Verasonics dataset, suggesting the potential to port features from high-end US systems to lower end counterparts.

关键词： Probes signal processing Array signal processing Filtering image quality image coding Recording Deep learning (DL) image quality in-phase and quadrature medical ultrasound imaging signal processing supervised learning

来源：评论

学校读者我要写书评

暂无评论

A fully automatic bearing fault diagnosis method based on an improved polar coordinate image texture

引用

MECHANICAL SYSTEMS AND signal processing 2025年 224卷

作者： Li, Bi Li, Zhinong He, Deqiang Guangxi Univ Sch Mech Engn Nanning 530004 Peoples R China Nanchang Hangkong Univ Key Lab Nondestruct Testing Minist Educ Nanchang 330063 Peoples R China

Previous researchers' Bearing Fault Diagnosis (BFD) methods often employ signal processing techniques to handle one-dimensional vibration signals, enabling the emergence of recognizable Bearing Fault Features (BFFs) in the Cartesian coordinate system. However, due to noise interference or limited by the manifestation of BFFs, these BFFs often require highly specialized personnel to identify and extract them, making the achievement of fully automated bearing fault diagnosis extremely challenging. Hence, a fully automatic BFD method based on Improved Polar Coordinate (IPC) image texture is proposed. Firstly, the proposed IPC algorithm transforms vibration signals into IPC images with easily recognizable BFFs in the polar coordinate system. Then, automatic image filtering, image texture enhancement, and texture feature extraction are achieved through methods in the field of image processing. Finally, automatic BFD experiments are conducted using extracted IPC image texture features and a neural network. The entire BFD process is fully automatic, and the methods employed are relatively simple and easy to implement, which is highly advantageous for promoting and implementing a real-time fault monitoring system. Experimental results show that the proposed fully automated BFD method based on IPC image texture is effective, achieving an average diagnostic accuracy of 99.4%. This surpasses the 95.0% accuracy of a similar method based on symmetrical polar coordinate image texture and the 98.9% accuracy of an advanced method based on refined composite multi-scale dispersion entropy. Moreover, the proposed method also has significant advantages in diagnosis efficiency compared to the advanced method.

关键词： Fully automatic BFD method IPC image texture image processing

来源：评论

学校读者我要写书评

暂无评论

An enhanced tree-seed algorithm for global optimization and neural architecture search optimization in medical image segmentation

引用

BIOMEDICAL signal processing AND CONTROL 2025年 104卷

作者： Qiao, Zenglin Wu, Lingyu Heidari, Ali Asghar Zhao, Xinchao Chen, Huiling Beijing Univ Posts & Telecommun Sch Sci Beijing 100876 Peoples R China Beijing Univ Posts & Telecommun Key Lab Math & Informat Networks Minist Educ Beijing Peoples R China Univ Tehran Coll Engn Sch Surveying & Geospatial Engn Tehran Iran Wenzhou Univ Dept Comp Sci & Artificial Intelligence Wenzhou 325035 Peoples R China

The spread of infectious diseases poses a threat to people's health. Screening and diagnosis using deep learning techniques can alleviate the pressure of the condition, especially medical image segmentation techniques, which can assist doctors inefficiently diagnosing and treating patients. However, most existing deep learning segmentation methods for medical images are mainly designed by experts based on their expertise. This paper proposes a novel wormhole and salp swarm strategy enhanced tree-seed algorithm (WSTSA). With its high efficiency, this algorithm could provide a sense of reassurance to the medical imaging field, instilling confidence in its potential. Secondly, WSTSA is integrated with a genetic algorithm to develop an automatic deep-learning neural architecture search model. Within this model, WSTSA optimizes hyperparameters during architecture search to enhance search accuracy, while the genetic algorithm explores the optimal convolutional neural network within a predefined search space. Finally, extensive experiments validate the performance of WSTSA and the proposed neural architecture search model. Statistical analyses demonstrate the superiority of WSTSA over existing state-of-the-art methods. Moreover, the neural architecture search model effectively discovers excellent neural networks for medical image segmentation.

关键词： Evolutionary deep learning Evolutionary algorithms image segmentation neural architecture search Tree-seed algorithm

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：