New learning subspace method for image feature extraction

Image pattern recognition is the most i mportant re-search directions currently,and feature extraction is itscoretechnology.Researchinthisfield dates backat leastto the 1940s ,but most feature extraction techniques re-quire users to give a priori probabil…     

基于核的最近邻特征重心分类器及人脸识别应用

本文提出核最近特征线和特征面分类器,可直接对高维人脸图像进行识别.为解决计算量大和可能失效的问题,提出(核)最近特征重心和(核)最近邻特征两种解决方法,前者降低了计算特征线和面距离的复杂度,后者减少了特征线和面的数目,两种方法均避免了可能失效的问题.将二者结合得到的(核)最近邻特征重心分类器,在获得相近识别率的条件下,使计算复杂度降到了最小.所得方法无需预先抽取人脸图像特征,因此避免了在较多样本数时,特征抽取存在计算量大的问题.基于ORL人脸数据库的实验验证了本文方法的有效性.     

Fast nearest neighbour testing algorithm for small feature sizes

A fast nearest neighbour algorithm with a logarithmic expected testing time for small feature sizes is presented. It has been tested on a robot vision application for YUV space. The algorithm time requirement is better than a multilayer perceptron trained for the same purpose.     

Generalised scheme for optimal learning in recurrent neuralnetworks

A new learning scheme is proposed for neural network architectures like the Hopfield network and bidirectional associative memory. This scheme, which replaces the commonly used learning rules, follows from the proof of the result that learning in these connectivity architectures is equivalent to learning in the 2-state perceptron. Consequently, optimal learning algorithms for the perceptron can be directly applied to learning in these connectivity architectures. Similar results are established for learning in the multistate perceptron, thereby leading to an optimal learning algorithm. Experimental results are provided to show the superiority of the proposed method     

Low-rank embedded orthogonal subspace learning for zero-shot classification

Zero-shot classification methods have attracted considerable attention in recent years. Existing ZSC methods encounter domain shift, hubness and visual–semantic gap problems. To address these problems, we propose a low-rank embedded orthogonal subspace learning method (LEOSL) for ZSC. Many previous works project visual features to the semantic space. However, they often suffer from the visual–semantic gap problem. To handle this problem, we project the visual representations and semantic representations to the common subspace. To address the domain shift problem, we restrict the mapping functions with a low-rank constraint. To handle the hubness problem, we introduce the class similarity term so that samples of the same class are located near each other, while samples of different classes are located far away. Furthermore, we restrict the shared representations in the subspace with an orthogonal constraint to remove the correlation between samples. The results show the superiority of LEOSL compared to many state-of-the-art methods.     

适用于组合特征识别的最近邻模糊分类器

在用多种特征进行简单的串联组合识别时,不同特征具有不同的特征类型和衡量尺寸,针对串联组合特征的这种特点,提出了一种最近邻模糊分类器.该分类器首先把待识别目标的组合特征与训练模板中的组合特征样本一一进行比较,从而得到了一个特征差矩阵.提出用模糊分布函数在同类特征差之间进行处理,生成一个隶属度矩阵,然后用算术平均法对隶属度矩阵进行处理,并用最大隶属度准则来进行分类判决.识别框架表明最近邻模糊分类器对组合特征中的各种不同特征的特征类型和衡量尺寸没有一致性要求,也无需对串联组合特征矢量做任何预处理.最后,用外场实测数据进行验证,结果表明,最近邻模糊分类器能够有效地解决多种特征串联组合的雷达目标识别问题.     

Generalised distributions of nonuniformity of hypergeometric transmission line

This letter outlines a general procedure for deriving the generalised expressions for nonuniformity patterns of all classes of solvable transmission lines, taking a class of hyper-geometric lines as an example for illustrating the application of the procedure.     

无监督行人重识别的判别性特征研究

目前,在计算机视觉方面,大多的监督学习方法用于解决其重要分支:行人重识别问题已经取得了不错的成果,但是此类方法需要对训练数据进行手工标注,特别是对于大容量的数据集,手工标注的成本很高,而且完全满足成对标记的数据难以获得,所以无监督学习成为必选项.此外,全局特征注重行人特征空间整体性的判别性,而局部特征有助于凸显不同部位特征的判别性.所以,基于全局与局部特征的无监督学习框架,使用全局损失函数与局部相斥损失函数共同进行判别性特征学习,并联合优化ResNet-50卷积神经网络(CNN)和各个样本之间的关系,最终实现行人重识别.大量实验数据验证了提出的方法在解决行人重识别任务时具有优越性.     

Beyond ITQ: Efficient binary multi-view subspace learning for instance retrieval

The existing hashing methods mainly handle either the feature based nearest-neighbor search or the category-level image retrieval, whereas a few efforts are devoted to instance retrieval problem. In this paper, we propose a binary multi-view fusion framework for directly recovering a latent Hamming subspace from the multi-view features for instance retrieval. More specifically, the multi-view subspace reconstruction and the binary quantization are integrated in a unified framework so as to minimize the discrepancy between the original multi-view high-dimensional Euclidean space and the resulting compact Hamming subspace. Besides, our method is essentially an unsupervised learning scheme without any labeled data involved, and thus can be used in the cases when the supervised information is unavailable or insufficient. Experiments on public benchmark and large-scale datasets reveal that our method achieves competitive retrieval performance comparable to the state-of-the-arts and has excellent scalability in large-scale scenario.     

仿射不变子空间特征及其在图像匹配中的应用

提出了一种新的基于子空间的仿射不变图像匹配算法。首先,分析了仿射不变特征空间的非线性几何结构,给出了图像特征的仿射不变子空间表示模型。其次,根据目标特征矩阵由它的正交投影矩阵唯一确定这一特性实现特征的全局一致性匹配。最后,通过计算同一目标在仿射变换下的子空间几何距离度量达到匹配识别的目的。模拟变换匹配实验和真实图像序列实验结果表明:所提出的匹配方法对仿射变化下的图像目标匹配具有很好的适应性和较高的鲁棒性。     

Generalised theory of a non-TEM transmission line taper without discontinuities

The optimum Chebyshev dispersive tapered transmission line requires impedance steps at the taper ends, which limits its application. A generalised theory for the synthesis of a transmission line taper without impedance steps and supporting non-TEM modes is presented. The impedance discontinuities are avoided by using a near-optimum frequency response.<>     

说话人语音特征子空间分离及识别应用

本文依据主元分析原理从语音特征观察空间分离说话人语音特征子空间,对输入语音特征矢量与子空间的距离测度进行了定义,并对基于特征子空间的说话人识别性能进行了分析.说话人语音训练样本提取特征后在语音特征观察空间形成具有一定散度的分布,根据主元分析原理和分布散度提取主要散度本征向量作为基底构成说话人语音特征子空间,并通过测试语音特征矢量与子空间的距离测度进行模式匹配.实验结果表明,特征子空间方法对说话人识别是有效的,特别是在小于3秒的短时测试语音下能够得到较高的识别率.     

A robust elastic net approach for feature learning

Unsupervised feature learning has drawn more and more attention especially in visual representation in past years. Traditional feature learning approaches assume that there are few noises in training data set, and the number of samples is enough compared with the dimensions of samples. Unfortunately, these assumptions are violated in most of visual representation scenarios. In these cases, many feature learning approaches are failed to extract the important features. Toward this end, we propose a Robust Elastic Net (REN) approach to handle these problems. Our contributions are twofold. First of all, a novel feature learning approach is proposed to extract features by weighting elastic net. A distribution induced weight function is used to leverage the importance of different samples thus reducing the effects of outliers. Moreover, the REN feature learning approach can handle High Dimension, Low Sample Size (HDLSS) issues. Second, a REN classifier is proposed for object recognition, and can be used for generic visual representation including that from the REN feature extraction. By doing so, we can reduce the effect of outliers in samples. We validate the proposed REN feature learning and classifier on face recognition and background reconstruction. The experimental results showed the robustness of this proposed approach for both corrupted/occluded samples and HDLSS issues.     

Transfer subspace learning based on structure preservation for JPEG image mismatched steganalysis

In real-world steganalysis applications, the traditional steganalysis methods built by a set of training data coming from a source may be applied to detect data from another new different source. In this case, the steganalyzers will face a serious problem that training data and test data are no longer subjected to the same distribution, and thus the detection performance would degrade rapidly. To address this problem, a novel transfer subspace learning method with structure preservation for image steganalysis is proposed in this paper. It aims to alleviate the mismatch between the training and test data so as to improve the detection performance. Specifically, a discriminant projection matrix is learned for the training and test data such that the projected data of both sets lie in a common subspace where each sample can be linearly reconstructed by a combination of the training data. In this way, the difference between the training and test sets is decreased. Further, in order to preserve the structure information of features in the projection subspace, a Frobenius-norm based regularization term is introduced into the objective function. Moreover, to mitigate the negative impacts of noises and outliers, a structurally sparse error matrix is introduced to model the noise and outlier information. The formulation of the proposed method can be efficiently solved by an alternating optimization algorithm. The extensive experiments compared with prior arts show the validity of the proposed method for JPEG image mismatched steganalysis.     

Kernel propagation strategy: A novel out-of-sample propagation projection for subspace learning

Kernel matrix optimization (KMO) aims at learning appropriate kernel matrices by solving a certain optimization problem rather than using empirical kernel functions. Since KMO is difficult to compute out-of-sample projections for kernel subspace learning, we propose a kernel propagation strategy (KPS) based on data distribution similar principle to effectively extract out-of-sample low-dimensional features for subspace learning with KMO. With KPS, we further present an example algorithm, i.e., kernel propagation canonical correlation analysis (KPCCA), which naturally fuses semi-supervised kernel matrix learning and canonical correlation analysis by means of kernel propagation projections. In KPCCA, the extracted correlation features of out-of-sample data not only incorporate integral data distribution information but also supervised information. Extensive experimental results have demonstrated the superior performance of our proposed method.     

Tunable line spectral estimators based on state-covariance subspace analysis

Subspace methods for spectral analysis can be adapted to the case where state covariance of a linear filter replaces the traditional Toeplitz matrix formed out of a partial autocorrelation sequence of a time series. This observation forms the basis of a new framework for spectral analysis. The goal of this paper is to quantify potential advantages in working with state-covariance data instead of the autocorrelation sequence. To this end, we identify tradeoffs between resolution and robustness in spectral estimates and how these are affected by the filter dynamics. The approach leads to a novel tunable high-resolution frequency estimator.