针对单一冗余字典在稀疏表示图像超分辨率重建结果出现不清晰、伪影以及重建过程编码效率不高、运算时间过长的问题,提出一种基于多字典学习和图像块映射的超分辨率重建方法。该方法在传统稀疏表示的框架下,首先探索局部图像块的梯度结...针对单一冗余字典在稀疏表示图像超分辨率重建结果出现不清晰、伪影以及重建过程编码效率不高、运算时间过长的问题,提出一种基于多字典学习和图像块映射的超分辨率重建方法。该方法在传统稀疏表示的框架下,首先探索局部图像块的梯度结构信息,按梯度角度将训练样本块分类;然后为每个子类样本集学习高低分辨率字典对,再结合最近邻思想应用生成的字典,为每个子类计算从低分辨率块到高分辨率块映射的函数;最后将重建过程简化为输入块和映射函数的乘积,在保证提高重建质量的同时减少了图像重建的时间。实验结果表明,所提算法在视觉效果有较大的提升,同时与锚点邻域回归算法相比,评价参数峰值信噪比(PSNR)平均提高约0.4 d B。展开更多
目的现有的基于邻域嵌入的人脸超分辨率重建算法只利用了低分辨率图像流形空间的几何结构,而忽略了原始高分辨率图像的流形几何结构,不能很好的反映高低分辨率图像流形几何结构的关系。此外,其对同一幅图像中的不同图像块选取固定数目...目的现有的基于邻域嵌入的人脸超分辨率重建算法只利用了低分辨率图像流形空间的几何结构,而忽略了原始高分辨率图像的流形几何结构,不能很好的反映高低分辨率图像流形几何结构的关系。此外,其对同一幅图像中的不同图像块选取固定数目的最近邻域图像块,从而导致重建质量的下降。为了充分利用原始高分辨率图像空间的几何结构信息,提出基于联合局部约束和自适应邻域选择的邻域嵌入人脸超分辨率重建算法。方法该方法结合待重构图像与低分辨率图像样本库的相似性约束与初始高分辨图像与高分辨率图像样本库的相似性约束,形成约束低分辨率图像块的重构权重,并利用该重构权重估计出高分辨率的人脸图像,同时引入自适应邻域选择的方法。结果在CAS-PEAL-R1人脸库上的实验结果表明,相较于传统的基于邻域嵌入的人脸超分辨率重建方法,本文算法在PSNR和SSIM上分别提升了0.39 d B和0.02。相较于LSR重建方法,在PSNR和SSIM上分别提升了0.63 d B和0.01;相较于LcR重建方法,在PSNR和SSIM上分别提升了0.36 d B和0.003 2;相较于TRNR重建方法,在PSNR和SSIM上分别提升了0.33 d B和0.001 1。结论本文所提的重建方法在现有人脸数据库上进行实验,在主观视觉和客观评价指标上均取得了较好的结果,可进一步适用于现实监控视频中人脸图像的高分辨率重建。展开更多
When high-impedance faults(HIFs)occur in resonant grounded distribution networks,the current that flows is extremely weak,and the noise interference caused by the distribution network operation and the sampling error ...When high-impedance faults(HIFs)occur in resonant grounded distribution networks,the current that flows is extremely weak,and the noise interference caused by the distribution network operation and the sampling error of the measurement devices further masks the fault characteristics.Consequently,locating a fault section with high sensitivity is difficult.Unlike existing technologies,this study presents a novel fault feature identification framework that addresses this issue.The framework includes three key steps:(1)utilizing the variable mode decomposition(VMD)method to denoise the fault transient zero-sequence current(TZSC);(2)employing a manifold learning algorithm based on t-distributed stochastic neighbor embedding(t-SNE)to further reduce the redundant information of the TZSC after denoising and to visualize fault information in high-dimensional 2D space;and(3)classifying the signal of each measurement point based on the fuzzy clustering method and combining the network topology structure to determine the fault section location.Numerical simulations and field testing confirm that the proposed method accurately detects the fault location,even under the influence of strong noise interference.展开更多
In phase space reconstruction of time series, the selection of embedding dimension is important. Based on the idea of checking the behavior of near neighbors in the reconstruction dimension, a new method to determine ...In phase space reconstruction of time series, the selection of embedding dimension is important. Based on the idea of checking the behavior of near neighbors in the reconstruction dimension, a new method to determine proper minimum embedding dimension is constructed. This method has a sound theoretical basis and can lead to good result. It can indicate the noise level in the data to be reconstructed, and estimate the reconstruction quality. It is applied to speech signal reconstruction and the generic embedding dimension of speech signals is deduced.展开更多
In this paper,we propose a refined local learning scheme to reconstruct a high resolution(HR)face image from a low resolution(LR)observation.The contribution of this work is twofold.Firstly,multi-direction gradient fe...In this paper,we propose a refined local learning scheme to reconstruct a high resolution(HR)face image from a low resolution(LR)observation.The contribution of this work is twofold.Firstly,multi-direction gradient features are extracted to search the nearest neighbors for each image patch,then the non-negative matrix factorization(NMF)is used to reduce the complexity in weight calculation,and the initial HR embedding is estimated from the training pairs by preserving local geometry.Secondly,a global reconstruction constraint and post-processing by non-local filtering is incorporated into super-resolution(SR)reconstruction process to reduce the image artifacts and further improve the image visual quality.Experimental results show that the proposed algorithm improves the SR performance both in subjective and objective assessments compared with several existing methods.展开更多
文摘针对单一冗余字典在稀疏表示图像超分辨率重建结果出现不清晰、伪影以及重建过程编码效率不高、运算时间过长的问题,提出一种基于多字典学习和图像块映射的超分辨率重建方法。该方法在传统稀疏表示的框架下,首先探索局部图像块的梯度结构信息,按梯度角度将训练样本块分类;然后为每个子类样本集学习高低分辨率字典对,再结合最近邻思想应用生成的字典,为每个子类计算从低分辨率块到高分辨率块映射的函数;最后将重建过程简化为输入块和映射函数的乘积,在保证提高重建质量的同时减少了图像重建的时间。实验结果表明,所提算法在视觉效果有较大的提升,同时与锚点邻域回归算法相比,评价参数峰值信噪比(PSNR)平均提高约0.4 d B。
文摘目的现有的基于邻域嵌入的人脸超分辨率重建算法只利用了低分辨率图像流形空间的几何结构,而忽略了原始高分辨率图像的流形几何结构,不能很好的反映高低分辨率图像流形几何结构的关系。此外,其对同一幅图像中的不同图像块选取固定数目的最近邻域图像块,从而导致重建质量的下降。为了充分利用原始高分辨率图像空间的几何结构信息,提出基于联合局部约束和自适应邻域选择的邻域嵌入人脸超分辨率重建算法。方法该方法结合待重构图像与低分辨率图像样本库的相似性约束与初始高分辨图像与高分辨率图像样本库的相似性约束,形成约束低分辨率图像块的重构权重,并利用该重构权重估计出高分辨率的人脸图像,同时引入自适应邻域选择的方法。结果在CAS-PEAL-R1人脸库上的实验结果表明,相较于传统的基于邻域嵌入的人脸超分辨率重建方法,本文算法在PSNR和SSIM上分别提升了0.39 d B和0.02。相较于LSR重建方法,在PSNR和SSIM上分别提升了0.63 d B和0.01;相较于LcR重建方法,在PSNR和SSIM上分别提升了0.36 d B和0.003 2;相较于TRNR重建方法,在PSNR和SSIM上分别提升了0.33 d B和0.001 1。结论本文所提的重建方法在现有人脸数据库上进行实验,在主观视觉和客观评价指标上均取得了较好的结果,可进一步适用于现实监控视频中人脸图像的高分辨率重建。
基金supported in part by the Science and Technology Program of State Grid Corporation of China(No.5108-202218280A-2-75-XG)the Fundamental Research Funds for the Central Universities(No.B200203129)the Postgraduate Research and Practice Innovation Program of Jiangsu Province(No.KYCX20_0432)。
文摘When high-impedance faults(HIFs)occur in resonant grounded distribution networks,the current that flows is extremely weak,and the noise interference caused by the distribution network operation and the sampling error of the measurement devices further masks the fault characteristics.Consequently,locating a fault section with high sensitivity is difficult.Unlike existing technologies,this study presents a novel fault feature identification framework that addresses this issue.The framework includes three key steps:(1)utilizing the variable mode decomposition(VMD)method to denoise the fault transient zero-sequence current(TZSC);(2)employing a manifold learning algorithm based on t-distributed stochastic neighbor embedding(t-SNE)to further reduce the redundant information of the TZSC after denoising and to visualize fault information in high-dimensional 2D space;and(3)classifying the signal of each measurement point based on the fuzzy clustering method and combining the network topology structure to determine the fault section location.Numerical simulations and field testing confirm that the proposed method accurately detects the fault location,even under the influence of strong noise interference.
基金Supported by the Naltural Science Foundation of Hunan Province(97JJY1006)Open Foundation of Stalte Key Lab. of Theory and Chief Technology on ISN of Xidian University(991894102)
文摘In phase space reconstruction of time series, the selection of embedding dimension is important. Based on the idea of checking the behavior of near neighbors in the reconstruction dimension, a new method to determine proper minimum embedding dimension is constructed. This method has a sound theoretical basis and can lead to good result. It can indicate the noise level in the data to be reconstructed, and estimate the reconstruction quality. It is applied to speech signal reconstruction and the generic embedding dimension of speech signals is deduced.
基金the National Natural Science Foundation of China(Nos.61171165 and 60802039)the Natural Science Foundation of Jiangsu(No.BK2010488)+1 种基金the Qing Lan Project of Jiangsu Province"the Six Top Talents"of Jiangsu Province Grant(No.2012DZXX-36)
文摘In this paper,we propose a refined local learning scheme to reconstruct a high resolution(HR)face image from a low resolution(LR)observation.The contribution of this work is twofold.Firstly,multi-direction gradient features are extracted to search the nearest neighbors for each image patch,then the non-negative matrix factorization(NMF)is used to reduce the complexity in weight calculation,and the initial HR embedding is estimated from the training pairs by preserving local geometry.Secondly,a global reconstruction constraint and post-processing by non-local filtering is incorporated into super-resolution(SR)reconstruction process to reduce the image artifacts and further improve the image visual quality.Experimental results show that the proposed algorithm improves the SR performance both in subjective and objective assessments compared with several existing methods.
