结合稀疏先验与多模式分解的低秩张量恢复方法

目的 各类终端设备获取的大量数据往往由于信息丢失而导致数据不完整,或经常受到降质问题的困扰。为有效恢复缺损或降质数据,低秩张量补全备受关注。张量分解可有效挖掘张量数据的内在特征,但传统分解方法诱导的张量秩函数无法探索张量不同模式之间的相关性;另外,传统张量补全方法通常将全变分约束施加于整体张量数据,无法充分利用张量低维子空间的平滑先验。为解决以上两个问题,提出了基于稀疏先验与多模式张量分解的低秩张量恢复方法。方法 在张量秩最小化模型基础上,融入多模式张量分解技术以及分解因子局部稀疏性。首先对原始张量施加核范数约束,以此捕获张量的全局低秩性,然后,利用多模式张量分解将整体张量沿着每个模式分解为一组低维张量和一组因子矩阵,以探索不同模式之间的相关性,对因子矩阵施加因子梯度稀疏正则化约束,探索张量子空间的局部稀疏性,进一步提高张量恢复性能。结果 在高光谱图像、多光谱图像、YUV（也称为YCbCr）视频和医学影像数据上,将本文方法与其他8种修复方法在3种丢失率下进行定量及定性比较。在恢复4种类型张量数据方面,本文方法与深度学习GP-WLRR方法（global prior refined weighted low-rank representation）的修复效果基本持平,本文方法的MPSNR（mean peak signal-to-noise ratio）在所有丢失率及张量数据上的总体平均高0.68dB,MSSIM（mean structural similarity）总体平均高0.01;与其他6种张量建模方法相比,本文方法的MPSNR及MSSIM均取得最优结果。结论 提出的基于稀疏先验与多模式张量分解的低秩张量恢复方法,可同时利用张量的全局低秩性与局部稀疏性,能够对受损的多维视觉数据进行有效修复。     

基于频率加权张量核范数的高光谱图像复原

目的 高光谱图像复原是高光谱领域中一个重要的预处理步骤,能够有效去除成像条件所带来的不利影响,提升后续处理任务的精度。张量核范数被广泛应用于高光谱复原问题中,得到了较好的结果。然而,在张量核范数的定义中,它对张量所有奇异值使用相同的阈值进行收缩,未充分考虑高光谱的物理意义,得到了次优的结果。为了提升高光谱图像复原的精度,本文提出了基于频率加权张量核范数的高光谱复原算法。方法 在张量的频率域内,对清晰的高光谱图像添加噪声,图像信息在低频部分变化较小,而在高频部分变化巨大。基于这样的物理意义,定义了一种频率加权张量核范数来逼近张量秩函数,提出了频率域权重的自适应确定方法,让其能减少对低频部分的收缩,同时加大高频部分惩罚。然后将其应用于高光谱图像复原和去噪问题中,并基于交替方向乘子法设计了相应最小化问题的快速求解算法。结果 在4个高光谱数据集上与相关方法进行对比仿真实验,高采样率条件下在Washington DC Mall数据集上,相比性能第2的模型,本文模型复原结果的PSNR （peak signal-to-noise ratio）提升了1.76 dB;在Stuff数据集上,PSNR值提升了2.91 dB。高噪声条件下,在Pavia数据集上相比性能第2的模型,本文模型去噪结果的PSNR提升了8.61 dB;在Indian数据集上,PSNR值提升了10.77 dB。结论 本文模型可以更好地探索高光谱图像的低秩特性,使复原的图像在保持主体信息的同时,复原出更多图像纹理细节。     

基于张量奇异值理论的交通数据重构方法

由于探测器和通信设备的故障,交通数据的缺失是不可避免的,这种缺失给智能交通系统（ITS）带来了不利的影响。针对此问题,运用张量平均秩的概念,对张量核范数进行最小化,从而构建了新的低秩张量补全模型,并且在此基础上,基于张量奇异值分解（T-SVD）和阈值分解（TSVT）理论,分别使用坐标梯度下降法（CGD）和交替乘子法（ADMM）对模型进行求解,提出两个张量补全算法LRTC-CGD和LRTC-TSVT。在公开的真实时空交通数据集上进行实验。结果表明,LRTC-CGD和LRTC-TSVT算法在不同的缺失场景和缺失率条件下,补全精度要优于现行的其他补全算法,并且在数据极端缺失情况下（70%~80%）,补全的效果更加稳定。     

Low-rank tensor completion with fractional-Jacobian-extended tensor regularization for multi-component visual data inpainting

Several low-rank tensor completion methods have been integrated with total variation (TV) regularization to retain edge information and promote piecewise smoothness. In this paper, we first construct a fractional Jacobian matrix to nonlocally couple the structural correlations across components and propose a fractional-Jacobian-extended tensor regularization model, whose energy functional was designed proportional to the mixed norm of the fractional Jacobian matrix. Consistent regularization could thereby be performed on each component, avoiding band-by-band TV regularization and enabling effective handling of the contaminated fine-grained and complex details due to the introduction of a fractional differential. Since the proposed spatial regularization is linear convex, we further produced a novel fractional generalization of the classical primal-dual resolvent to develop its solver efficiently. We then combined the proposed tensor regularization model with low-rank constraints for tensor completion and addressed the problem by employing the augmented Lagrange multiplier method, which provides a splitting scheme. Several experiments were conducted to illustrate the performance of the proposed method for RGB and multispectral image restoration, especially its abilities to recover complex structures and the details of multi-component visual data effectively.     

A Mixture of Nuclear Norm and Matrix Factorization for Tensor Completion

In this paper, we propose a mixture model for tensor completion by combining the nuclear norm with the low-rank matrix factorization. To solve this model, we develop two algorithms: non-smooth low-rank tensor completion (NS-LRTC), smooth low-rank tensor completion (S-LRTC). When the sampling rate (SR) is high, our experiments on real-world data show that the NS-LRTC algorithm outperforms other tested methods in running time and recovery quality. In addition, whatever the SR is, the proposed S-LRTC algorithm delivers state-of-art recovery performance compared with other tested approaches. Although the objective function in our model is non-convex and non-differentiable, we prove that every cluster point of the sequence generated by NS-LRTC or S-LRTC is a stationary point.     

面向多视图聚类的低秩张量表示学习

针对现有鲁棒图学习忽略多视图间的互补信息和高阶相关性问题,提出一种面向多视图聚类的低秩张量表示学习（LRTRL-MVC）算法。利用鲁棒主成分分析的思想,在去除噪声的干净数据上计算各视图的鲁棒图和转移概率矩阵,然后构建一个包含各视图马尔可夫转移概率矩阵的张量,采用基于张量奇异值分解的核范数来确保目标张量的低秩性质。利用迭代最优化算法求解,将求得的低秩张量作为马尔可夫谱聚类算法的输入得到最终聚类结果。在4个不同类型的公开标准数据集BBCSport、NGs、Yale和MSRCv1上进行实验并与相关的最好多视图聚类算法进行对比,结果表明在3个聚类度量标准下,所提算法的聚类结果均高于其他对比算法。     

基于非局部张量火车分解的彩色图像修补

数据在采集和转换的过程中通常存在部分数据丢失的问题,丢失数据的补全直接影响后续的识别、跟踪等高层任务的结果.自然图像中经常存在许多具有重复特性的相似结构,利用该类冗余信息,文中提出基于非局部张量火车分解的张量补全方法.利用图像的非局部相似性,挖掘其中蕴含的低秩特性,并通过张量火车分解模型进行建模及升阶,将低阶张量转化为高阶以进行低秩信息的进一步挖掘利用,从而进行图像中缺失数据的修补.实验验证文中方法在图像修补上的有效性.     

基于lP范数的非凸低秩张量最小化

在低秩矩阵、张量最小化问题中,凸函数容易求得最优解,而非凸函数可以得到更低秩的局部解.文中基于非凸替换函数的低秩张量恢复问题,提出基于l_p范数的非凸张量模型.采用迭代加权核范数算法求解模型,实现低秩张量最小化.在合成数据和真实图像上的大量实验验证文中方法的恢复性能.     

基于非凸低秩约束的图像修复方法

受传输干扰或存储不当等因素的影响,现实应用中获取的某些图像通常会存在像素缺失现象,这给图像的后续分析与处理带来了一定影响.解决该问题的常用方法是对图像进行低秩修复.利用低秩特性进行修复的方法大多以秩函数建模,由于矩阵秩函数是非凸离散的,该模型的求解是一个NP难问题,所以通常利用核范数对矩阵的秩进行凸松弛.但是,基于核范...     

截断核范数和全变差正则化高光谱图像复原

目的 高光谱图像距具有较高的光谱分辨率,从而具备区分诊断性光谱特征地物的能力,但高光谱数据经常会受到如环境、设备等各种因素的干扰,导致数据污染,严重影响高光谱数据在应用中的精度和可信度。方法 根据高光谱图像光谱维度特征值大小与所包含信息的关系,利用截断核范数最小化方法表示光谱低秩先验,从而有效抑制稀疏噪声;再利用高光谱图像的空间稀疏先验建立正则化模型,达到去除高密度噪声的目的;最终,结合上述两种模型的优势,构建截断核范数全变差正则化模型去除高斯噪声、稀疏噪声及其他混合噪声等。结果 将本文与其他三种近期发表的主流去噪方法进行对比,模型平均峰信噪比提高3.20 dB,平均结构相似数值指标提高0.22,并可以应用到包含各种噪声、不同尺寸的图像,其模型平均峰信噪比提高1.33 dB。结论 本文方法在光谱低秩中更加准确地表示了观测数据的先验特征,利用高光谱遥感数据的空间和低秩先验信息,能够对含有高密度噪声以及稀疏异常值的图像进行复原。     

基于准范数与零范数的矩阵恢复

针对高维度矩阵的低秩恢复问题中核范数与l1范数过惩罚导致的结果偏差,提出一种矩阵恢复方法。使用准范数代替低秩恢复问题中常用的核范数约束,使用零范数代替l1范数约束。对于准范数的求解问题,采用与准范数等价的Frobenius/核混合范数进行替代,提出基于交替近似的线性最小化方法对目标函数进行求解。在合成数据与真实数据上的实验结果表明,该算法在主观视觉效果与客观数值比较上都能获得更好的结果。     

Weighted Nuclear Norm Minimization and Its Applications to Low Level Vision

As a convex relaxation of the rank minimization model, the nuclear norm minimization (NNM) problem has been attracting significant research interest in recent years. The standard NNM regularizes each singular value equally, composing an easily calculated convex norm. However, this restricts its capability and flexibility in dealing with many practical problems, where the singular values have clear physical meanings and should be treated differently. In this paper we study the weighted nuclear norm minimization (WNNM) problem, which adaptively assigns weights on different singular values. As the key step of solving general WNNM models, the theoretical properties of the weighted nuclear norm proximal (WNNP) operator are investigated. Albeit nonconvex, we prove that WNNP is equivalent to a standard quadratic programming problem with linear constrains, which facilitates solving the original problem with off-the-shelf convex optimization solvers. In particular, when the weights are sorted in a non-descending order, its optimal solution can be easily obtained in closed-form. With WNNP, the solving strategies for multiple extensions of WNNM, including robust PCA and matrix completion, can be readily constructed under the alternating direction method of multipliers paradigm. Furthermore, inspired by the reweighted sparse coding scheme, we present an automatic weight setting method, which greatly facilitates the practical implementation of WNNM. The proposed WNNM methods achieve state-of-the-art performance in typical low level vision tasks, including image denoising, background subtraction and image inpainting.     

基于张量核范数与3D全变分的背景减除

针对常用背景减除方法忽略前景时空连续性的问题,以及动态背景对前景提取的干扰问题,基于张量鲁棒主成分分析（TRPCA）提出了一种改进的背景减除模型。该模型利用改进的张量核范数对背景进行约束,加强了背景的低秩性,保留了视频的空间信息;然后用3D全变分（3D-TV）对前景进行正则化约束,考虑了目标在时空上的连续性,有效地抑制了动态背景和目标移动对前景提取造成的干扰。实验结果表明,所提算法能有效地分离视频中的前景和背景,且与高阶鲁棒主成分分析（HoRPCA）、带有新核范数的张量鲁棒主成分分析（TRPCA-TNN）和基于克罗内克基的鲁棒主成分分析（KBR-RPCA）等方法相比,综合评判指标F-measure值均处于最优或次优状态。由此可见,所提算法有效地提高了前景背景分离的准确度,抑制了复杂天气和目标移动对前景提取的干扰。     

Low Rank Prior and Total Variation Regularization for Image Deblurring

The similar image patches should have similar underlying structures. Thus the matrix constructed from stacking the similar patches together has low rank. Based on this fact, the nuclear norm minimization, which is the convex relaxation of low rank minimization, leads to good denoising results. Recently, the weighted nuclear norm minimization has been applied to image denoising. This approach presents state-of-the-art result for image denoising. In this paper, we further study the weighted nuclear norm minimization problem for general image recovery task. For the weights being in arbitrary order, we prove that such minimization problem has a unique global optimal solution in the closed form. Incorporating this idea with the celebrated total variation regularization, we then investigate the image deblurring problem. Numerical experimental results illustratively clearly that the proposed algorithms achieve competitive performance.     

基于加权核范数最小化的矩阵填充模型

协同过滤是目前推荐系统最常用的技术之一,相比于传统的推荐技术具有一定优势,但其缺点是受用户对商品评价的稀疏性制约,现阶段一般利用矩阵填充技术来解决这一问题。主要研究了基于低秩的矩阵填充模型,针对原有模型解对所有奇异值用同一值收缩的问题,提出了一种加权核范数最小化模型以提高核范数灵活度,给出了该模型用收缩算子可得到全局最优解的相关定理及证明,同时对模型的另一种形式在求解过程中的迭代收敛性进行了证明。用凸优化主流算法在两种真实数据集上进行的实验表明,改进后的模型一定程度上提高了计算速度与准确性。     

基于张量核范数与3D全变分的背景减除

针对常用背景减除方法忽略前景时空连续性的问题，以及动态背景对前景提取的干扰问题，基于张量鲁棒主成分分析（TRPCA）提出了一种改进的背景减除模型。该模型利用改进的张量核范数对背景进行约束，加强了背景的低秩性，保留了视频的空间信息；然后用3D全变分（3D-TV）对前景进行正则化约束，考虑了目标在时空上的连续性，有效地抑制了动态背景和目标移动对前景提取造成的干扰。实验结果表明，所提算法能有效地分离视频中的前景和背景，且与高阶鲁棒主成分分析（HoRPCA）、带有新核范数的张量鲁棒主成分分析（TRPCA-TNN）和基于克罗内克基的鲁棒主成分分析（KBR-RPCA）等方法相比，综合评判指标F-measure值均处于最优或次优状态。由此可见，所提算法有效地提高了前景背景分离的准确度，抑制了复杂天气和目标移动对前景提取的干扰。     

Tensor factorization using auxiliary information

Most of the existing analysis methods for tensors (or multi-way arrays) only assume that tensors to be completed are of low rank. However, for example, when they are applied to tensor completion problems, their prediction accuracy tends to be significantly worse when only a limited number of entries are observed. In this paper, we propose to use relationships among data as auxiliary information in addition to the low-rank assumption to improve the quality of tensor decomposition. We introduce two regularization approaches using graph Laplacians induced from the relationships, one for moderately sparse cases and the other for extremely sparse cases. We also give present two kinds of iterative algorithms for approximate solutions: one based on an EM-like algorithms which is stable but not so scalable, and the other based on gradient-based optimization which is applicable to large scale datasets. Numerical experiments on tensor completion using synthetic and benchmark datasets show that the use of auxiliary information improves completion accuracy over the existing methods based only on the low-rank assumption, especially when observations are sparse.     

非凸加权核范数及其在运动目标检测中的应用

目的 近年来,低秩矩阵分解被越来越多的应用到运动目标检测中。但该类方法一般将矩阵秩函数松弛为矩阵核函数优化,导致背景恢复精度不高;并且没有考虑到前景目标的先验知识,即区域连续性。为此提出一种结合非凸加权核范数和前景目标区域连续性的目标检测算法。方法 本文提出的运动目标检测模型以鲁棒主成分分析(RPCA)作为基础,在该基础上采用矩阵非凸核范数取代传统的核范数逼近矩阵低秩约束,并结合了前景目标区域连续性的先验知识。该方法恢复出的低秩矩阵即为背景图像矩阵,而稀疏大噪声矩阵则是前景目标位置矩阵。结果 无论是在仿真数据集还是在真实数据集上,本文方法都能够取得比其他低秩类方法更好的效果。在不同数据集上,该方法相对于RPCA方法,前景目标检测性能提升25%左右,背景恢复误差降低0.5左右;而相对于DECOLOR方法,前景目标检测性能提升约2%左右,背景恢复误差降低0.2左右。结论 矩阵秩函数的非凸松弛能够比凸松弛更准确的表征出低秩特征,从而在运动目标检测应用中更准确的恢复出背景。前景目标的区域连续性先验知识能够有效地过滤掉非目标大噪声产生的影响,使得较运动目标检测的精度得到大幅提高。因此,本文方法在动态纹理背景、光照渐变等较复杂场景中均能够较精确地检测出运动目标区域。但由于区域连续性的要求,本文方法对于小区域多目标的检测效果不甚理想。     

A Bayesian robust CP decomposition approach for missing traffic data imputation

The inevitable problem of missing data is ubiquitous in the real transportation system, which makes the data-driven intelligent transportation system suffer from incorrect response. We propose a Bayesian robust Candecomp/Parafac (CP) tensor decomposition (BRCP) approach to deal with missing data and outliers by integrating the general form of transportation system domain knowledge. Specifically, when the lower rank tensor captures the global information, the sparse tensor is added to capture the local information, which can robustly predict the distribution of missing items and under the fully Bayesian treatment, the effective variational reasoning can prevent the over fitting problem. Real and reliable traffic data sets are used to evaluate the performance of the model in two data missing scenarios, which the experimental results show that the proposed BRCP model achieves the best imputation accuracy and is better than the most advanced baseline (Bayesian Gaussian CP decomposition (BGCP), high accuracy low-rank tensor completion (HaLRTC) and SVD-combined tensor decomposition (STD)), even in the case of high missed detection rate, the model still has the best performance and robustness.