Extended linear regression for undersampled face recognition

Linear Regression Classification (LRC) is a newly-appeared pattern recognition method, which formulates the recognition problem in terms of class-specific linear regression with sufficient training samples per class. In this paper, we extend LRC via intraclass variant dictionary and SVD to undersampled face recognition where there are very few, or even only one, training sample per class. Intraclass variant dictionary is adopted in undersampled situation to represent the possible variation between the training and testing samples. Three types of methods, quasi-inverse, ridge regularization and Singular Value Decomposition (SVD), are designed to solve low-rank problem of data matrix. Then the whole algorithm, named Extended LRC (ELRC), is presented for face recognition via intraclass variant dictionary and SVD. The experimental results on three well-known face databases show that the proposed ELRC has better generalization ability and is more robust to classification than many state-of-the-art methods in undersampled situation.     

FLDA方法在单样本人脸识别中的应用研究

在人脸识别应用中,当每个人有多个训练样本(MSPP)时,Fisher线性判别分析(FLDA)方法可以很好地用于特征提取.然而,当每个人只有一个训练样本(SSPP)时,因为类内散布矩阵为零矩阵,所以FLDA方法将不能使用.为了解决该问题,提出了一种比较新颖的方法来估计类内散布矩阵,借助于奇异值分解(SVD)方法,先将人脸图像分解成两部分,然后分别估计出类内散布矩阵及类间散布矩阵,使FLDA方法能够得到有效的应用.在ORL及Yale上的实验表明了提出的方法比现有的许多方法取得了更好的识别效果.     

基于酉变换-矩阵束的稀布线阵方向图综合

该文提出一种非迭代的稀布线阵方向图综合方法。该方法首先对方向图采样数据进行centro-Hermit化处理,然后通过酉变换构造等价实矩阵束,得到非均匀单元位置与新矩阵束广义特征值的关系。在此基础上,对实矩阵奇异值分解,并舍弃非主要奇异值以获得低阶左奇异向量矩阵,进而求得稀布阵列的阵元位置和相应激励。相比于其他方法,该方法能够直接得到阵元位置的实数解,奇异值分解和特征值分解均在实数域进行,提高逼近程度的同时有效降低了计算量,仿真验证了该方法利用少量阵元即可高效实现线阵的方向图综合。     

基于判别性降维的字典学习在人脸识别中的应用

针对目前大多数人脸识别方法只能单独实施降维或者字典学习而不能完全利用训练样本判别信息的问题,提出了基于判别性降维的字典学习方法,通过联合降维与字典学习使得投影矩阵和字典更好地相互拟合,从而可以获得更高效的人脸分类系统。所提方法的有效性在AR及MPIE两大通用人脸数据库上得到了验证,实验结果表明,相比于几种先进的线性表示方法,所提算法取得了更高的识别率,特别当训练样本数很少的时候,识别效果更佳。     

字典学习优化判别性降维的鲁棒人脸识别

针对现有的人脸识别算法由于光照、表情、姿态、面部遮挡等变化而严重影响识别性能的问题,提出了基于字典学习优化判别性降维的鲁棒人脸识别算法。首先,利用经典的特征提取算法PCA初始化降维投影矩阵;然后,计算字典和系数,通过联合降维与字典学习使得投影矩阵和字典更好地相互拟合;最后,利用迭代算法输出字典和投影矩阵,并利用经l2-范数正则化的分类器完成人脸的识别。在扩展YaleB、AR及一个户外人脸数据库上的实验验证了本文算法的有效性及鲁棒性,实验结果表明,相比几种线性表示算法,本文算法在处理鲁棒人脸识别时取得了更高的识别率。     

Robust zero-watermark algorithms based on numerical relationship between adjacent blocks

In this paper, three robust zero-watermark algorithms named Direct Current coefficient RElationship (DC-RE), CUmulant combined Singular Value Decomposition (CU-SVD), and CUmulant combined Singular Value Decomposition RElationship (CU-SVD-RE) are proposed. The algorithm DC-RE gets the feature vector from the relationship of DC coefficients between adjacent blocks, CU-SVD gets the feature vector from the singular value of third-order cumulants, while CU-SVD-RE combines the essence of the first two algorithms. Specially, CU-SVD-RE gets the feature vector from the relationship between singular values of third-order cumulants. Being a cross-over studying field of watermarking and cryptography, the zero-watermark algorithms are robust without modifying the carrier. Numerical simulation obviously shows that, under geometric attacks, the performance of CU-SVD-RE and DC-RE algorithm are better and all three proposed algorithms are robust to various attacks, such as median filter, salt and pepper noise, and Gaussian low-pass filter attacks.     

Singular Value Decomposition Rank-Deficient-Based Estimators in TD-SCDMA Systems

This paper presents a new joint channel estimation method for Time Division-Synchronous Code Division Multiple Access (TD-SCDMA) systems based on a Singular Value Decomposition (SVD) reduced-rank technique. The system capacity is increased by increasing the highest number of users in one time slot. The additional channel estimation processing required for the increasing number of users is solved by adopting the reduced-rank technique, which estimates a limited number of parameters that are needed to describe the channel matrix and reduce the dimensionality of this matrix. Simulation results prove the validity of the proposed reduced-rank technique for channel estimation accuracy enhancing. Additionally, it is shown that the detectors based on the reduced-rank estimators outperform traditional channel estimators, contributing to 7.8 and 6.9 dB BER performance improvements for indoor and vehicular channels, respectively.     

采用奇异值分解方法的半盲信道辨识算法

针对全盲信道辨识算法无法辨识含公零点信道且对信道阶数误差敏感的问题,提出一种半盲信道辨识算法。通过奇异值分解将信道矩阵分解为同维矩阵与酉矩阵乘积的形式,分别利用接收数据和已知符号求解同维矩阵与酉矩阵,最终得到信道矩阵的闭式解。该算法有效地克服了全盲信道辨识算法的诸多局限性,避免了传统半盲方法面临的最优加权选择问题,性能稳定,且对信道噪声与信道阶数都具有较强的鲁棒性。仿真实验验证了所提算法的有效性与优越性。     

High-accuracy band to band registration method for multi-spectral images of HJ-1A/B

Band-to-band registration accuracy is an important parameter of multispectral data. A novel band-to-band registration approach with high precision is proposed for the multi-spectral images of HJ-1A/B. Firstly, the main causes resulted in misregistration are analyzed, and a high-order polynomial model is proposed. Secondly, a phase fringe filtering technique is employed to Phase Correlation Method based on Singular Value Decomposition (SVD-PCM) for reducing the noise in phase difference matrix. Then, experiments are carried out to build nonlinear registration models, and images of green band and red band are aligned to blue band with an accuracy of 0.1 pixels, while near infrared band with an accuracy of 0.2 pixels.     

Sparse representation for robust face recognition by dictionary decomposition

Sparse representation-based classification (SRC) method has gained great success in face recognition due to its encouraging and impressive performance. However, in SRC the data used to train or test are usually corrupted, and hence the performance is affected. This paper proposes a robust face recognition approach by means of learning a class-specific dictionary and a projection matrix. Firstly, the training data are decomposed into class-specific dictionary, non-class-specific dictionary, and sparse error matrix. Secondly, in order to correct the corrupted test data, the data are projected onto their corresponding underlying subspace, and a projection matrix between the original training data and the class-specific dictionary is learned. Then, the features of the class-specific dictionary and the corrected test data are extracted by using Eigenface method. Finally, the SRC is performed to classify. Extensive experiments conducted on publicly available data sets show that the proposed algorithm performs better than some state-of-the-art methods.     

基于局部约束编码的稀疏保持投影降维识别方法研究

稀疏表示技术的引入可有效解决降维处理对图参数的依赖,但这类降维方法不能同时兼顾稀疏重构和样本数据的邻近性问题。针对该问题,本文提出了一种基于局部约束编码的稀疏保持投影降维识别方法。通过稀疏表示分类模型构建了图边权矩阵,引入局部约束因子设计了降维投影模型,推导降维求解过程,分析了本文方法与SPP （ Sparse Preserving Projections ）和SLPP（ Soft Locality Preserving Projections ）方法之间的共性和区别,最后给出了识别算法流程。采用人脸图像数据集和高分辨SAR（ Synthetic Aperture Radar ）图像数据集对算法的有效性进行仿真验证,由于考虑了数据间的邻近性,本文方法较传统方法可获得更好的识别性能。     

应用改进的主要特征基函数快速计算目标宽角度RCS

特征基函数法是分析目标宽角度电磁散射特性的有效方法之一,但在构造特征基函数时,设置的入射波激励包含大量的冗余信息,大大降低了特征基函数的构造效率;另外在分析复杂目标时,在增加激励数目的情况下,仅应用主要特征基函数并不能显著提高计算精度。针对这些问题,该文对特征基函数构造方法进行改进,首先采用奇异值分解技术对激励矩阵进行压缩去除冗余信息,减少求解矩阵方程的次数;其次充分考虑子域之间的互耦作用,将主要特征基函数与次要特征基函数融合,得到改进的主要特征基函数。数值计算结果表明:与传统方法相比,该方法具有更高的计算效率和计算精度。     

基于四元数的Root-MUSIC的双基地MIMO雷达中角度估计算法

该文将四元数理论应用到双基地集中式多输入多输出(MIMO)雷达的角度估计中。文中通过传统数据模型构造四元数矩阵,提出了基于四元数的求根-多重信号分类(Root MUltiple SIgnal Classification, Root-MUSIC)的MIMO雷达中角度估计算法,该算法通过奇异值分解和Root-MUSIC来估计出发射角(Direction Of Departure, DOD)和接收角(Direction Of Arrival, DOA)。该算法的角度估计性能远优于现有文献的方法,并且无需谱峰搜索,复杂度大大降低。仿真结果验证了算法的有效性。     

矩阵填充及其在信号处理中的应用

本文首先阐述了矩阵填充的应用背景,给出了矩阵填充的数学模型,详细分析了矩阵填充中的低秩特性和非相干特性,重点介绍了矩阵填充三种典型的重构算法:SVT(Singular Value Thresholding)算法、ADMiRA(Atomic Decomposition for Minimum Rank Approximation)算法和SVP(Singular Value Projection)算法,文中的仿真实验对这三种算法的重构性能进行了比较;文章随后分析了矩阵填充和压缩感知的联系;最后介绍了矩阵填充在协同过滤、系统识别、传感器网络、图像处理、稀疏信道估计、频谱感知以及多媒体编码和通信等方面的的应用。      

基于奇异值分解的低速率波形内插语音编码算法

波形内插(WI)语音编码模型作为当今最具潜力的低速率语音编码方案之一,因其良好的性能,越来越受到人们的重视.本文基于一种奇异值分解(SVD)的特征波形分解方法,利用语音信号的感知特性,将二维特征波形的幅度谱分成基本矩阵、过渡矩阵和补充矩阵,并采用了不同的量化方法,有效地降低了运算复杂度;另外,本文根据语音信号时变特性,将三个矩阵分为三种组合模式表示特征波形幅度谱,并引入周期因子和能量熵来衡量矩阵周期程度,解决了奇异值分解后参数难于量化的问题,提高了编码效率.主观A/B测试表明,本文提出的2.4kbps SVD-WI编码器的重建语音质量略好于2.4kbps MELP编码器.     

基于GSVD的核不相关辨别子空间与雷达目标识别

该文提出了一种基于广义奇异值分解的核不相关辨别子空间算法,并将其用于高分辨距离像雷达目标识别。新算法结合广义奇异值分解与核方法的优点,有效地解决了传统方法面临的矩阵奇异问题,同时进一步改善了目标的类可分性。其次,依据Fisher准则导出了距离像总散度矩阵零空间中不含有有用辨别信息的结论。利用这一结论,可以在求解核不相关最优辨别矢量之前对各散度矩阵进行预降维,以减小后续运算的计算复杂度。对3类飞机目标实测数据的识别结果表明了所提方法的有效性。     

基于平行互质虚拟阵列的低复杂度二维DOA联合估计算法

针对传统平行阵列2维测向自由度低、分辨能力差和小快拍情况下估计误差大等问题,该文提出基于平行互质虚拟阵列的低复杂度2维波达角(DOA)估计算法.该算法利用两个相互平行的互质线阵扩展生成虚拟阵列,并通过协方差矩阵和互协方差矩阵构造具有增强2维角度自由度的扩展矩阵,最后通过奇异值分解(SVD)和旋转不变技术(ESPRIT)...     

基于优化的LDA算法人脸识别研究

提取低维人脸特征是人脸识别系统中极其关键的一步。线性判别分析(LDA)是一种较为普遍的用于特征提取的线性分类方法。本文提出了一种优化的LDA算法,该方法克服了传统的LDA算法用于人脸识别时存在的问题:通过重新定义样本类间离散度矩阵使传统的Fisher准则能够最优化,克服了边缘类对选择最佳投影方向的影响;同时,利用因数分解的方法避免了对矩阵求逆,解决了小样本问题。依据经验选取适当的e值,得到最佳的识别效果。实验结果表明,人脸识别效果优于传统LDA方法。     

Interpolation of head-related transfer functions using spherical fourier expansion

A new interpolation algorithm for Head-Related Transfer Function (HRTF) is proposed to realize 3D sound reproduction via headphones in arbitrary spatial direction. HRTFs are modeled as a weighted sum of spherical harmonics on a spherical surface. Truncated Singular Value Decomposition (SVD) is adopted to calculate the weights of the model. The truncation number is chosen according to Frobenius norm ratio and the partial condition number. Compared with other interpolated methods, our proposed approach not on...     

基于奇异值分解的小波域水印算法

结合奇异值分解（SVD）和离散小渡变换（DWT）的特点,提出一种基于SVD的小波域数字图像水印算法。该算法将二值水印图像经过取反置乱后嵌入到原始图像小波中频子带的奇异值中,具有较高的抗攻击能力。仿真实验证明,该算法不仅具有良好的透明性,而且对常见攻击,如：叠加噪声、JPEG压缩、滤波及几何攻击具有较好的鲁棒性。