期刊文献+
共找到1,379篇文章
< 1 2 69 >
每页显示 20 50 100
A Physical Security Technology Based upon Doubly Multiple Parameters Weighted Fractional Fourier Transform
1
作者 Li Yong Sun Teng +2 位作者 Sha Xuejun Song Zhiqun Wang Bin 《China Communications》 SCIE CSCD 2024年第10期200-209,共10页
Enhancing the security of the wireless communication is necessary to guarantee the reliable of the data transmission, due to the broadcast nature of wireless channels. In this paper, we provide a novel technology refe... Enhancing the security of the wireless communication is necessary to guarantee the reliable of the data transmission, due to the broadcast nature of wireless channels. In this paper, we provide a novel technology referred to as doubly multiple parameters weighted fractional Fourier transform(DMWFRFT), which can strengthen the physical layer security of wireless communication. This paper introduces the concept of DM-WFRFT based on multiple parameters WFRFT(MP-WFRFT), and then presents its four properties. Based on these properties, the parameters decryption probability is analyzed in terms of the number of parameters. The number of parameters for DM-WFRFT is more than that of the MP-WFRFT,which indicates that the proposed scheme can further strengthen the the physical layer security. Lastly, some numerical simulations are carried out to illustrate that the efficiency of proposed DM-WFRFT is related to preventing eavesdropping, and the effect of parameters variety on the system performance is associated with the bit error ratio(BER). 展开更多
关键词 doubly multiple parameters weighted fractional Fourier transform(DM-WFRFT) physical layer security transform parameters variety
在线阅读 下载PDF
基于Transformer的异类目标智能关联跟踪
2
作者 鹿瑶 李子豪 +1 位作者 刘准钆 杨衍波 《航空学报》 北大核心 2025年第17期267-282,共16页
考虑异类多目标跟踪,现有跟踪系统模型参数固定,未能充分挖掘异类目标运动特性差异并将其体现在模型参数上,从而造成跟踪精度下降、甚至出现失跟或误跟现象。针对这一问题,提出一种基于Transformer的异类目标智能关联跟踪算法。在时间... 考虑异类多目标跟踪,现有跟踪系统模型参数固定,未能充分挖掘异类目标运动特性差异并将其体现在模型参数上,从而造成跟踪精度下降、甚至出现失跟或误跟现象。针对这一问题,提出一种基于Transformer的异类目标智能关联跟踪算法。在时间滑窗下,结合稀疏注意力和自注意力机制,提取历史雷达点迹数据/短时航迹数据中的目标类型特征、点迹航迹关联特征,动态调整系统过程噪声协方差矩阵进行异类目标精细化运动建模与雷达点迹/目标航迹的一对一可靠关联,以获得高质量异类目标跟踪输出。不同杂波密度下异类目标跟踪典型场景仿真结果表明,相较于联合概率数据关联滤波、多假设跟踪、概率假设密度滤波和模型参数固定的智能关联跟踪方法,所提方法具有更高的跟踪精度和稳定性。 展开更多
关键词 异类多目标跟踪 数据关联 短时航迹分类 过程噪声协方差动态调整 transformer网络
原文传递
基于Transformer和多序列特征的土壤含水率预测
3
作者 邝晓飞 万里平 +3 位作者 连嘉茜 段欣玥 尉鹏亮 郭交 《农业机械学报》 北大核心 2025年第8期120-127,共8页
土壤含水率(Soil moisture content,SMC)的准确预测在农业生产中至关重要。多时间序列多源遥感数据可提供多种特征的时间变化信息,但多时相多序列信息往往在SMC反演中得不到有效利用。Transformer网络在处理多序列特征中表现优越,本研... 土壤含水率(Soil moisture content,SMC)的准确预测在农业生产中至关重要。多时间序列多源遥感数据可提供多种特征的时间变化信息,但多时相多序列信息往往在SMC反演中得不到有效利用。Transformer网络在处理多序列特征中表现优越,本研究基于Transformer结构构建了提取SMC的深度回归模型,并将其与卷积神经网络回归(Convolutional neural network regression,CNNR)、长短期记忆网络(Long short-term memory,LSTM)回归、门控循环单元(Gated recurrent unit,GRU)回归进行比较。实验结果表明,使用长时间序列特征数据更有利于SMC预测;在利用5 d的历史数据预测5 d后的SMC时,Transformer回归相较于CNNR、LSTM和GRU,决定系数平均提升0.0953、0.0324、0.0336,均方根误差平均降低0.014、0.0026、0.0030 cm^(3)/cm^(3)。对特征影响和中间隐藏特征的变化分析显示,为不同时刻特征分配合适的注意力更有利于预测SMC。 展开更多
关键词 土壤含水率 模型分析 多时间序列 transformER 多源遥感
在线阅读 下载PDF
Experimental and mechanistic study on iron extraction from high-iron red mud under multiple physical field coupling conditions
4
作者 DONG Hai-pei YANG Jin-lin +3 位作者 ZHOU Wen-tao YU Xu-yang MA Shao-jian WANG Ding-zheng 《Journal of Central South University》 2025年第7期2476-2486,共11页
Red mud is a solid waste discharged in the process of alumina production,and how to realize the efficient recovery of its iron is an urgent problem to be solved.In this study,the iron extraction test and mechanism stu... Red mud is a solid waste discharged in the process of alumina production,and how to realize the efficient recovery of its iron is an urgent problem to be solved.In this study,the iron extraction test and mechanism study of high iron red mud were carried out under the coupling conditions of multiple physical field(microwave field,gas-solid flow field and temperature field)with biomass as the reducing agent.The test results showed that under the optimal conditions,an iron concentrate with a yield of 78.4%,an iron grade of 59.23%,and a recovery rate of 86.65%was obtained.The analyses of XRD,XPS,TEM,and SEM-EDS showed that during the roasting process,the hematite in the high-iron red mud was completely converted to magnetite,and the biomass produced the reductant that provided the magnetization reaction;A large number of cracks and pores appeared in the surface of the hematite reduction product particles,which helped to induce iron minerals to undergo effective mineral phase transformation.The above study provides ideas for the phase transformation and efficient recovery of iron minerals in red mud. 展开更多
关键词 multiple physical field high-iron red mud BIOMASS mineral phase transformation iron extraction
在线阅读 下载PDF
基于Transformer的轴承故障诊断
5
作者 王磊 张宇 《河北冶金》 2025年第1期41-46,共6页
轴承是旋转机械的重要零件,其健康状态直接影响到工业设备的运行稳定性。同时,由于轴承本身结构特性和工作环境等原因,极易发生损伤和损坏。因此,对轴承状态进行有效的故障诊断,对于保障设备和人员安全性、减少维修成本、降低停机时间... 轴承是旋转机械的重要零件,其健康状态直接影响到工业设备的运行稳定性。同时,由于轴承本身结构特性和工作环境等原因,极易发生损伤和损坏。因此,对轴承状态进行有效的故障诊断,对于保障设备和人员安全性、减少维修成本、降低停机时间具有重要的意义。尽管,基于数据驱动的深度学习模型在轴承故障诊断领域取得了巨大的成功,但大多数是基于不变工况下的,意味着训练数据和测试数据必须遵从相同或相似的分布。然而,机械设备通常需要在不同工况下运行,训练数据和测试数据会产生分布偏移,导致模型的测试诊断精度显著下降,严重限制了基于深度学习故障诊断模型的应用。为此,提出了一种针对变工况条件的轴承故障诊断模型——E-Transformer。该模型首先利用信号嵌入完成对一维信号的分割和高维表示,在高维空间里丰富信息,然后通过多头注意力机制对嵌入的子信号进一步处理,寻找信号在不同空间里的状态特征并深度表征。经凯斯西储大学轴承数据集验证,该模型诊断准确率优于其他几种先进的深度学习方法。 展开更多
关键词 E-transformer 轴承 故障诊断 变工况 多头注意力机制 信号表征
在线阅读 下载PDF
融合CNN与Transformer的MRI脑肿瘤图像分割 被引量:6
6
作者 刘万军 姜岚 +2 位作者 曲海成 王晓娜 崔衡 《智能系统学报》 CSCD 北大核心 2024年第4期1007-1015,共9页
为解决卷积神经网络(convolutional neural network,CNN)在学习全局上下文信息和边缘细节方面受到很大限制的问题,提出一种同时学习局语义信息和局部空间细节的级联神经网络用于脑肿瘤医学图像分割。首先将输入体素分别送入CNN和Transfo... 为解决卷积神经网络(convolutional neural network,CNN)在学习全局上下文信息和边缘细节方面受到很大限制的问题,提出一种同时学习局语义信息和局部空间细节的级联神经网络用于脑肿瘤医学图像分割。首先将输入体素分别送入CNN和Transformer分支,在编码阶段结束后,采用一种双分支融合模块将2个分支学习到的特征有效地结合起来以实现全局信息与局部信息的融合。双分支融合模块利用哈达玛积对双分支特征之间的细粒度交互进行建模,同时使用多重注意力机制充分提取特征图通道和空间信息并抑制无效的噪声信息。在BraTS竞赛官网评估了本文方法,在BraTS2019验证集上增强型肿瘤区、全肿瘤区和肿瘤核心区的Dice分数分别为77.92%,89.20%和81.20%。相较于其他先进的三维医学图像分割方法,本文方法表现出了更好的分割性能,为临床医生做出准确的脑肿瘤细胞评估和治疗方案提供了可靠依据。 展开更多
关键词 医学图像分割 脑肿瘤 级联神经网络 卷积神经网络 transformER 特征融合 多重注意力 残差学习
在线阅读 下载PDF
Multiple attenuation using λ–f domain high-resolution Radon transform 被引量:7
7
作者 李志娜 李振春 +1 位作者 王鹏 徐强 《Applied Geophysics》 SCIE CSCD 2013年第4期433-441,511,共10页
The parabolic Radon transform has been widely used in multiple attenuation. To further improve the accuracy and efficiency of the Radon transform, we developed the 2- fdomain high-resolution Radon transform based on t... The parabolic Radon transform has been widely used in multiple attenuation. To further improve the accuracy and efficiency of the Radon transform, we developed the 2- fdomain high-resolution Radon transform based on the fast and modified parabolic Radon transform presented by Abbad. The introduction of a new variable 2 makes the transform operator frequency-independent. Thus, we need to calculate the transform operator and its inverse operator only once, which greatly improves the computational efficiency. Besides, because the primaries and multiples are distributed on straight lines with different slopes in the 2-fdomain, we can easily choose the filtering operator to suppress the multiples. At the same time, the proposed method offers the advantage of high-resolution Radon transform, which can greatly improve the precision of attenuating the multiples. Numerical experiments suggest that the multiples are well suppressed and the amplitude versus offset characteristics of the primaries are well maintained. Real data processing results further verify the effectiveness and feasibility of the method. 展开更多
关键词 λ-f domain high resolution parabolic Radon transform multiple attenuation
在线阅读 下载PDF
基于遥感多参数和CNN-Transformer的冬小麦单产估测 被引量:5
8
作者 王鹏新 杜江莉 +3 位作者 张悦 刘峻明 李红梅 王春梅 《农业机械学报》 EI CAS CSCD 北大核心 2024年第3期173-182,共10页
为了提高冬小麦单产估测精度,改善估产模型存在的高产低估和低产高估等现象,以陕西省关中平原为研究区域,选取旬尺度条件植被温度指数(VTCI)、叶面积指数(LAI)和光合有效辐射吸收比率(FPAR)为遥感特征参数,结合卷积神经网络(CNN)局部特... 为了提高冬小麦单产估测精度,改善估产模型存在的高产低估和低产高估等现象,以陕西省关中平原为研究区域,选取旬尺度条件植被温度指数(VTCI)、叶面积指数(LAI)和光合有效辐射吸收比率(FPAR)为遥感特征参数,结合卷积神经网络(CNN)局部特征提取能力和基于自注意力机制的Transformer网络的全局信息提取能力,构建CNN-Transformer深度学习模型,用于估测关中平原冬小麦产量。与Transformer模型(R^(2)为0.64,RMSE为465.40 kg/hm^(2),MAPE为8.04%)相比,CNN-Transformer模型具有更高的冬小麦单产估测精度(R^(2)为0.70,RMSE为420.39 kg/hm^(2),MAPE为7.65%),能够从遥感多参数中提取更多与产量相关的信息,且对于Transformer模型存在的高产低估和低产高估现象均有所改善。基于5折交叉验证法和留一法进一步验证了CNN-Transformer模型的鲁棒性和泛化能力。此外,基于CNN-Transformer模型捕获冬小麦生长过程的累积效应,分析逐步累积旬尺度输入参数对产量估测的影响,评估模型对于冬小麦不同生长阶段的累积过程的表征能力。结果表明,模型能有效捕捉冬小麦生长的关键时期,3月下旬至5月上旬是冬小麦生长的关键时期。 展开更多
关键词 冬小麦 作物估产 遥感多参数 卷积神经网络 transformer模型
在线阅读 下载PDF
Multiple Darboux–B?cklund transformations via truncated Painleve′ expansion and Lie point symmetry approach 被引量:4
9
作者 Shuai-Jun Liu Xiao-Yan Tang Sen-Yue Lou 《Chinese Physics B》 SCIE EI CAS CSCD 2018年第6期103-108,共6页
For a given truncated Painleve′ expansion of an arbitrary nonlinear Painleve′ integrable system, the residue with respect to the singularity manifold is known as a nonlocal symmetry, called the residual symmetry, wh... For a given truncated Painleve′ expansion of an arbitrary nonlinear Painleve′ integrable system, the residue with respect to the singularity manifold is known as a nonlocal symmetry, called the residual symmetry, which is proved to be localized to Lie point symmetries for suitable prolonged systems. Taking the Korteweg–de Vries equation as an example, the n-th binary Darboux–Ba¨cklund transformation is re-obtained by the Lie point symmetry approach accompanied by the localization of the n-fold residual symmetries. 展开更多
关键词 residue symmetry multiple Darboux-Baicklund transformation Lie point symmetry approach
原文传递
Multiple Time Scale Analysis of River Runoff Using Wavelet Transform for Dagujia River Basin, Yantai, China 被引量:10
10
作者 LIU Delin LILT Xianzhao +2 位作者 LI Bicheng ZHAO Shiwei LI Xiguo 《Chinese Geographical Science》 SCIE CSCD 2009年第2期158-167,共10页
Based on monthly river runoff and meteorological data, a method of Morlet wavelet transform was used to analyze the multiple time scale characteristics of river runoff in the Dagujia River Basin, Yantai City, Shandong... Based on monthly river runoff and meteorological data, a method of Morlet wavelet transform was used to analyze the multiple time scale characteristics of river runoff in the Dagujia River Basin, Yantai City, Shandong Province. The results showed that the total annual river runoffin the Dagujia River Basin decreased significantly from 1966 to 2004, and the rate of decrease was 48× 10^6ma/10yr, which was higher than the mean value of most rivers in China. Multiple time scale characteristics existed, which accounted for different aspects of the changes in annual river runoff, and the major periods of the runofftime series were identified as about 28 years, 14 years and 4 years with decreasing levels of fluctuation. The river runoff evolution process was controlled by changes in precipitation to a certain extent, but it was also greatly influenced by human activities. Also, for different time periods and scales, the impacts of climate changes and human activities on annual river runoff evolution occurred at the same time. Changes in the annual river runoffwere mainly associated with climate change before the 1980s and with human activities after 1981. 展开更多
关键词 multiple time scale river runoff climate change Morlet wavelet transform Dagujia River
在线阅读 下载PDF
An image encryption algorithm based on improved baker transformation and chaotic S-box 被引量:4
11
作者 Xing-Yuan Wang Huai-Huai Sun Hao Gao 《Chinese Physics B》 SCIE EI CAS CSCD 2021年第6期216-225,共10页
The algorithm is an image encryption algorithm based on the improved baker transformation and chaotic substitution box(S-box). It mainly uses the initial values and parameters of a one-dimensional logistic chaotic sys... The algorithm is an image encryption algorithm based on the improved baker transformation and chaotic substitution box(S-box). It mainly uses the initial values and parameters of a one-dimensional logistic chaotic system as an encryption key. Specifically, in the image scrambling stage, the algorithm primarily uses an improved baker transform method to process the image. In the image diffusion stage, the algorithm first uses the chaotic S-box method to process the encryption key. Secondly, an exclusive OR(XOR) operation is performed on the image and the encryption key to initially diffuse the image. Finally, the image is again diffused using the method of ortho XOR. Simulation analysis shows that the algorithm can achieve good encryption effect, simple and easy implementation, and good security. In the digital image communication transmission, it has good practical value. 展开更多
关键词 image encryption improved baker transformation chaotic s-box chaotic sequence
原文传递
Multiple description scalable video coding based on 3D lifted wavelet transform 被引量:3
12
作者 KIM Yong-deak 《Journal of Zhejiang University-Science A(Applied Physics & Engineering)》 SCIE EI CAS CSCD 2006年第5期857-863,共7页
In this work, a new method to deal with the unconnected pixels in motion compensated temporal filtering (MCTF) is presented, which is designed to improve the performance of 3D lifted wavelet coding. Furthermore, multi... In this work, a new method to deal with the unconnected pixels in motion compensated temporal filtering (MCTF) is presented, which is designed to improve the performance of 3D lifted wavelet coding. Furthermore, multiple description scalable coding (MDSC) is investigated, and novel MDSC schemes based on 3D wavelet coding are proposed, using the lifting imple- mentation of temporal filtering. The proposed MDSC schemes can avoid the mismatch problem in multiple description video coding, and have high scalability and robustness of video transmission. Experimental results showed that the proposed schemes are feasible and adequately effective. 展开更多
关键词 multiple DESCRIPTION SCALABLE coding (MDSC) MOTION compensated temporal filtering (MCTF) Block-split bidi-rectional MOTION estimation 3D lifted WAVELET transform
在线阅读 下载PDF
LNG-Transformer:基于多尺度信息交互的图像分类网络 被引量:3
13
作者 王文杰 杨燕 +2 位作者 敬丽丽 王杰 刘言 《计算机科学》 CSCD 北大核心 2024年第2期189-195,共7页
鉴于Transformer的Self-Attention机制具有优秀的表征能力,许多研究者提出了基于Self-Attention机制的图像处理模型,并取得了巨大成功。然而,基于Self-Attention的传统图像分类网络无法兼顾全局信息和计算复杂度,限制了Self-Attention... 鉴于Transformer的Self-Attention机制具有优秀的表征能力,许多研究者提出了基于Self-Attention机制的图像处理模型,并取得了巨大成功。然而,基于Self-Attention的传统图像分类网络无法兼顾全局信息和计算复杂度,限制了Self-Attention的广泛应用。文中提出了一种有效的、可扩展的注意力模块Local Neighbor Global Self-Attention(LNG-SA),该模块在任意时期都能进行局部信息、邻居信息和全局信息的交互。通过重复级联LNG-SA模块,设计了一个全新的网络,称为LNG-Transformer。该网络整体采用层次化结构,具有优秀的灵活性,其计算复杂度与图像分辨率呈线性关系。LNG-SA模块的特性使得LNG-Transformer即使在早期的高分辨率阶段,也可以进行局部信息、邻居信息和全局信息的交互,从而带来更高的效率、更强的学习能力。实验结果表明,LNG-Transformer在图像分类任务中具有良好的性能。 展开更多
关键词 图像分类 自注意力机制 多尺度 transformER
在线阅读 下载PDF
基于变量选择与Transformer模型的中长期电力负荷预测方法 被引量:12
14
作者 黄文琦 梁凌宇 +3 位作者 王鑫 赵翔宇 宗珂 孙凌云 《浙江大学学报(理学版)》 CAS CSCD 北大核心 2024年第4期483-491,500,共10页
准确且有效的负荷预测对于电力系统的实时运行和调度非常重要。提出了一种融合变量选择与稀疏Transformer模型的预测方法,将静态变量和时序变量作为输入,充分发挥静态变量在全局时间范围内的信息增强作用,基于门控机制设计变量分权组件... 准确且有效的负荷预测对于电力系统的实时运行和调度非常重要。提出了一种融合变量选择与稀疏Transformer模型的预测方法,将静态变量和时序变量作为输入,充分发挥静态变量在全局时间范围内的信息增强作用,基于门控机制设计变量分权组件,根据变量与预测结果的相关性,赋予变量不同的权重。设计了双层编码结构,进行时序特征提取,对注意力进行稀疏处理,通过多变量输入对未来时刻负荷进行预测。基于真实电力负荷数据的实验表明,本文模型能够提高中长期负荷预测精度和效率。 展开更多
关键词 电力时序数据 transformER 中长期负荷预测 多变量 变量选择
在线阅读 下载PDF
Multiple Image Watermarking with Discrete Cosine Transform 被引量:1
15
作者 Mahbuba Begum Mohammad Shorif Uddin 《Journal of Computer and Communications》 2021年第3期88-94,共7页
Multiple image watermarking is an advanced concept of singular watermarking techniques. The existing singular digital image watermarking techniques cannot obtain all the design goals, such as imperceptibility, robustn... Multiple image watermarking is an advanced concept of singular watermarking techniques. The existing singular digital image watermarking techniques cannot obtain all the design goals, such as imperceptibility, robustness, security, and capacity simultaneously with perfection. Hence, the multiple image watermarking technique is designed that embeds several watermarks into the same host image for conveying multiple information. This paper proposed a multiple image watermarking technique with Discrete Cosine Transform (DCT) for ensuring the low computational time for embedding, encryption, decryption as well as extraction of the watermark images. Here, DCT is used to ensure the visual quality of the host image, and a random binary matrix is used to improve the security of the digital image. We have given a basic framework for multiple image watermarking. Our experimental results show satisfactory performance. 展开更多
关键词 multiple Watermarking Discrete Cosine transform Discrete Wavelet transform Arnold transform
在线阅读 下载PDF
Magnetic-field-driven reverse martensitic transformation with multiple magneto-responsive effects by manipulating magnetic ordering in Fe-doped Co-V-Ga Heusler alloys 被引量:1
16
作者 Kai Liu Shengcan Ma +7 位作者 Yuxi Zhang Hai Zeng Guang Yu Xiaohua Luo Changcai Chen Sajjad Ur Rehman Yongfeng Hu Zhenchen Zhong 《Journal of Materials Science & Technology》 SCIE EI CAS CSCD 2020年第23期145-154,共10页
Nowadays,searching for the materials with multiple magneto-functional properties and good mechanical properties is vital in various fields,such as solid-state refrigeration,magnetic actuators,magnetic sensors and inte... Nowadays,searching for the materials with multiple magneto-functional properties and good mechanical properties is vital in various fields,such as solid-state refrigeration,magnetic actuators,magnetic sensors and intelligent/smart devices.In this work,the magnetic-field-induced metamagnetic reverse martensitic transformation(MFIRMT)from paramagnetic martensite to ferromagnetic austenite with multiple magneto-responsive effects is realized in Fe-doped Co-V-Ga Heusler alloys by manipulating the magnetic ordering.The martensitic transformation temperature Tmreduces quasi-linearly with increasing Fe-content.In strikingly contrast with the Fe-free alloys,the magnetization difference(M')across martensitic transformation increases by three orders of magnitude for Fe-doped alloys.The increased M'should be ascribed to the reduction of Tm,almost unchanged Curie temperature of austenite and the increased magnetic moment in the samples with higher Fe-content.The large M'provides strong driving force to realize the MFIRMT and accordingly multiple magneto-responsive effects,such as magnetocaloric,magnetoresistance and magnetostriction effects.Meanwhile,giant Vickers hardness of 518 HV and compressive strength of 1423 MPa are achieved.Multiple magneto-responsive effects with exceptional mechanical properties make these alloys great potential candidates for applications in many fields. 展开更多
关键词 Magneto-functional properties Mechanical properties Ferromagnetic ordering Reverse martensitic transformation multiple magneto-responsive effects Co-based Heusler alloys
原文传递
Boundary Element Analysis (Laplace Transform Solution) of Groundwater Unsteady Flow to a Multiple Well System in a Confined Aquifer 被引量:1
17
作者 Zhu Xueyu Xie Chunhong Zou Zeyuan Nanjing University, Nanjing, Jiangsu 《Acta Geologica Sinica(English Edition)》 SCIE CAS CSCD 1990年第1期93-99,共7页
The calculations of unsteady flow to a multiple well system with the application of boundary elementmethod (BEM) are discussed. The mathematical model of unsteady well flow is a boundary value problem ofparabolic diff... The calculations of unsteady flow to a multiple well system with the application of boundary elementmethod (BEM) are discussed. The mathematical model of unsteady well flow is a boundary value problem ofparabolic differential equation. It is changed into an elliptic one by Laplace transform to eliminate time varia-ble. The image function of water head H can be solved by BEM. We derived the boundary integral equation ofthe transformed variable H and the discretization form of it, so that there is no need to discretize the bounda-ries of well walls and it becomes easier to solve the groundwater head H by numerical inversion. 展开更多
关键词 of Groundwater Unsteady Flow to a multiple Well System in a Confined Aquifer Laplace transform Solution Boundary Element Analysis FLOW
在线阅读 下载PDF
基于Transformer增强型U-net的CT图像稀疏重建与伪影抑制 被引量:5
18
作者 樊雪林 文昱齐 乔志伟 《CT理论与应用研究(中英文)》 2024年第1期1-12,共12页
实现低剂量计算机断层成像(CT)的一个有效办法是减少投影角度,但投影角度较少会产生严重的条状伪影,降低图像的临床使用价值。针对该问题,提出一种耦合卷积神经网络(CNN)和多种注意力机制的U型网络(TE-unet)。首先采用U型架构提取多尺... 实现低剂量计算机断层成像(CT)的一个有效办法是减少投影角度,但投影角度较少会产生严重的条状伪影,降低图像的临床使用价值。针对该问题,提出一种耦合卷积神经网络(CNN)和多种注意力机制的U型网络(TE-unet)。首先采用U型架构提取多尺度特征信息;其次提出一个包含CNN和多种注意力的模块提取图像特征;最后在跳跃连接处加入Transformer块过滤信息,抑制不相关特征,突出重要特征。所提网络结合CNN的局部特征提取能力和Transformer的全局信息捕获能力,辅以多种注意力机制,实现了良好的去条状伪影能力。在60个投影角度下,与经典的Uformer网络相比,峰值信噪比(PSNR)高出0.3178 dB,结构相似度(SSIM)高出0.002,均方根误差(RMSE)降低0.0005。实验结果表明,所提TE-unet重建的图像精度更高,图像细节保留的更好,可以更好地压制条状伪影。 展开更多
关键词 稀疏重建 计算机断层成像 transformER 多注意力机制 条状伪影
原文传递
基于Transformer的无人机多目标跟踪算法研究 被引量:5
19
作者 苑玉彬 吴一全 《兵器装备工程学报》 CAS CSCD 北大核心 2024年第7期11-18,共8页
针对无人机多目标跟踪面临目标遮挡、尺度变化、快速运动、复杂环境等问题,提出了一种基于Transformer的无人机多目标跟踪算法,采用Focal Transformer来捕获高分辨率输入的Transformer层中的局部和全局交互。该算法能够生成目标的检测... 针对无人机多目标跟踪面临目标遮挡、尺度变化、快速运动、复杂环境等问题,提出了一种基于Transformer的无人机多目标跟踪算法,采用Focal Transformer来捕获高分辨率输入的Transformer层中的局部和全局交互。该算法能够生成目标的检测信息以及目标的外观特征,从而提高了跟踪性能。在轨迹预测方面,采用了卡尔曼滤波方法,以准确地预测目标的运动轨迹,有助于提高跟踪的准确性和鲁棒性。在数据关联过程中,同时考虑了检测置信度、外观嵌入距离和IOU距离3个因素,以更有效地处理数据关联过程,提高了多目标跟踪模型的鲁棒性,使其能够在复杂场景中更好地跟踪目标。此外,还使用了轨迹的二次匹配方法,进一步提高了算法的性能。在VisDrone和UAVDT数据集上进行了对比验证,证明了该算法在实际应用中的有效性和可行性。本研究为无人机多目标跟踪提供了一种新的解决方案,具有广泛的应用前景。 展开更多
关键词 多目标跟踪 transformER 卡尔曼滤波 检测置信度 多特征匹配
在线阅读 下载PDF
Joint Non-Orthogonal Multiple Access (NOMA) & Walsh-Hadamard Transform: Enhancing the Receiver Performance 被引量:2
20
作者 Muhammad Rehan Usman Arsla Khan +1 位作者 Muhammad Arslan Usman Soo young Shin 《China Communications》 SCIE CSCD 2018年第9期160-177,共18页
Non-orthogonal multiple access(NOMA) is a new access method to achieve high performance gains in terms of capacity and throughput, so it is currently under consideration as one of the candidates for fifth generation(5... Non-orthogonal multiple access(NOMA) is a new access method to achieve high performance gains in terms of capacity and throughput, so it is currently under consideration as one of the candidates for fifth generation(5 G) technologies. NOMA utilizes power domain in order to superimpose signals of multiple users in a single transmitted signal. This creates a lot of interference at the receive side. Although the use of successive interference cancellation(SIC) technique reduces the interference, but to further improve the receiver performance, in this paper, we have proposed a joint Walsh-Hadamard transform(WHT) and NOMA approach for achieving better performance gains than the conventional NOMA. WHT is a well-known code used in communication systems and is used as an orthogonal variable spreading factor(OVSF) in communication systems. Application of WHT to NOMA results in low bit error rate(BER) and high throughput performance for both low and high channel gain users. Further, it also reduces peak to average power ratio(PAPR) of the user signal. The results are discussed in terms of comparison between the conventionalNOMA and the proposed technique, which shows that it offers high performance gains in terms of low BER at different SNR levels, reduced PAPR, high user throughput performance and better spectral efficiency. 展开更多
关键词 Bit error rate (BER) non-orthogonal multiple access (NOMA) orthogonal frequency division multiplexing (OFDM) peak to average power ratio (PAPR) successive interference cancellation (SIC) throughput and Walsh-Hadamard transform (WHT).
在线阅读 下载PDF
上一页 1 2 69 下一页 到第
使用帮助 返回顶部