利用协方差矩阵法表示深度图像集的鲁棒人脸识别

针对深度图像集人脸识别的鲁棒性问题,提出将多幅Kinect图像作为一个图像集,Kinect捕获的原始深度数据可用于姿态估计以及人脸区域的自动裁剪。首先,将图像集划分到c个图像子集,子集中的所有图像划分为4×4的图像块;然后,将图像集中的图像模拟为图像块,按照姿势划分,每个子集使用协方差矩阵法表示;最后,在黎曼流形上模拟子集图像,为了分类,黎曼流形的每个图像子集分别学习支持向量机模型,并引入一种融合方法来合并所有图像子集的结果。在三个最大的公开Kinect人脸数据集Curtin Faces、Biwi Kinect和UWA Kinect上的实验结果验证了该方法的有效性,与其他较先进的方法相比,识别率有较大提升,标准差保持较低,对图像集数量、图像子集划分数量和空间分辨率具有较好的鲁棒性。

非平稳随机振动分析的协方差矩阵法

本文在平稳随机振动分析的协方差矩阵法基础上,建立了非平稳随机振动分析的协方差矩阵法。采用本法对受有随机激励的多自由度振动系统进行了分析计算,然后与理论稳态解进行了比较,结果表明协方差矩阵法用来解非平稳随机振动问题是可靠和有效的,它既可用于线性时不变系统,又可用于线性时变系统,只要能得到激励的均值或协方差矩阵,便可采用本法求得系统响应的均值和方差。

估算测量不确定度的最大熵法和协方差矩阵法

本文评述了测量不确定度估算方法的研究动向和新进展，提出并论述了估算测量不确定度的两种新方法一最大熵法和协方差矩阵法．最大熵法能够对未知的误差概率分布做出主观偏见最小的最佳估计。协方差矩阵法，评定不确定度的模型通用性好，应用范围广，误差概率分布可以是未知的，采用矩阵运算，算法简洁，适用于计算机快速处理测量数据。

联合协方差特征矩的时空上下文鲁棒视觉跟踪算法

时空上下文(Spatio-temporal context,STC)算法利用上下文中包含的有用信息提升目标跟踪的鲁棒性,取得了良好的跟踪效果。本文联合协方差特征矩的算法增强目标跟踪的鲁棒性,以STC算法跟踪结果置信值(RC)作为算法选择条件,根据RC判断是否用协方差特征矩的算法进行跟踪校正,自适应目标尺度大小,用Meanshift算法确定目标中心位置。最后在公共数据集上进行的对比实验表明,所提出的算法具有更好的跟踪效果和鲁棒性。

三种传感器布点优化算法在系杆拱桥中的对比分析

对三种不同的传感器优化布点算法在系杆拱桥中的应用进行对比分析,研究不同算法的侧重点及优化效果。首先对三种优化方法的原理进行分析,然后以一大跨系杆拱桥为背景,研究桥面在三种方法下加速度传感器测点布设的差异,并采用四种评估准则对传感器布置的优化效果进行对比分析,最后采用无量纲化技术将四种评估准则转化为综合评价体系。分析结果表明:不同的优化算法对优化效果的影响较大,在进行传感器的优化布设时,必须对不同的优化算法进行对比,以选取最优算法。

风险值 (VaR) 度量方法新进展 下

认真作好风险度量和管理工作,保持金融机构的稳健经营,是现代经济运行的基石。近年来,为了更好地衡量资产损失的风险,人们提出了风险值(Value at Risk)的概念,目前,风险值(VaR)已经成为风险管理目标的同义词。本文讨论的是风险值度量方法的新进展。具体包括三部分内容:第一部分是对风险管理概念VaR的讨论,指出国内一些文章对VaR概念的一些不恰当理解和应用;第二部分是完善VaR度量方法的分类标准及名称,介绍了风险值度量方法的新进展,并给出了VaR度量方法的实施程序;第三部分是对风险值度量方法研究的展望。

Robust range-parameterized cubature Kalman filter for bearings-only tracking

In order to improve tracking accuracy when initial estimate is inaccurate or outliers exist,a bearings-only tracking approach called the robust range-parameterized cubature Kalman filter(RRPCKF)was proposed.Firstly,the robust extremal rule based on the pollution distribution was introduced to the cubature Kalman filter(CKF)framework.The improved Turkey weight function was subsequently constructed to identify the outliers whose weights were reduced by establishing equivalent innovation covariance matrix in the CKF.Furthermore,the improved range-parameterize(RP)strategy which divides the filter into some weighted robust CKFs each with a different initial estimate was utilized to solve the fuzzy initial estimation problem efficiently.Simulations show that the result of the RRPCKF is more accurate and more robust whether outliers exist or not,whereas that of the conventional algorithms becomes distorted seriously when outliers appear.

An Ocean Reanalysis System for the Joining Area of Asia and Indian-Pacific Ocean

An ocean reanalysis system for the joining area of Asia and Indian-Pacific Ocean (AIPO) has been developed and is currently delivering reanalysis data sets for study on the air-sea interaction over AIPO and its climate variation over China in the inter-annual time scale.This system consists of a nested ocean model forced by atmospheric reanalysis,an ensemble-based multivariate ocean data assimilation system and various ocean observations.The following report describes the main components of the data assimilation system in detail.The system adopts an ensemble optimal interpolation scheme that uses a seasonal update from a free running model to estimate the background error covariance matrix.In view of the systematic biases in some observation systems,some treatments were performed on the observations before the assimilation.A coarse resolution reanalysis dataset from the system is preliminarily evaluated to demonstrate the performance of the system for the period 1992 to 2006 by comparing this dataset with other observations or reanalysis data.

Geomagnetic jerk extraction based on the covariance matrix

We normalize data from 43 Chinese observatories and select data from ten Chinese observatories with most continuous records to assess the secular variations(SVs)and geomagnetic jerks by calculating the deviations between annual observed and CHAOS-6 model monthly means.The variations in the north,east,and vertical eigendirections are studied by using the covariance matrix of the residuals,and we find that the vertical direction is strongly affected by magnetospheric ring currents.To obtain noise-free data,we rely on the covariance matrix of the residuals to remove the noise contributions from the largest eigenvalue or vectors owing to ring currents.Finally,we compare the data from the ten Chinese observatories to seven European observatories.Clearly,the covariance matrix method can simulate the SVs of Dst,the jerk of the northward component in 2014 and that of the eastward component in 2003.5 in China are highly agree with that of Vertically downward component in Europe,compare to CHAOS-6,covariance matrix method can show more details of SVs.

Interference Alignment Based on Subspace Tracking in MIMO Cognitive Networks with Multiple Primary Users

The interference alignment （IA） algorithm based on FDPM subspace tracking （FDPM-ST IA） is proposed for MIMO cognitive network （CRN） with multiple primary users in this paper. The feasibility conditions of FDPM-ST IA is also got. Futherly, IA scheme of secondary network and IA scheme of primary network are given respectively without assuming a priori knowledge of interference covariance matrices. Moreover, the paper analyses the computational complexity of FDPM-ST IA. Simulation results and theoretical calculations show that the proposed algorithm can achieve higher sum rate with lower computational complexity.

The REMO Ocean Data Assimilation System into HYCOM(RODAS_H):General Description and Preliminary Results

The first version of the Brazilian Oceano- graphic Modeling and Observation Network （REMO） ocean data assimilation system into the Hybrid Coordi- nate Ocean Model （HYCOM） （RODAS H） has recently been constructed for research and operational purposes. The system is based on a multivariate Ensemble Optimal Interpolation （EnOI） scheme and considers the high fre- quency variability of the model error co-variance matrix. The EnOl can assimilate sea surface temperature （SST）, satellite along-track and gridded sea level anomalies （SLA）, and vertical profiles of temperature （T） and salinity （S） from Argo. The first observing system experiment was carried out over the Atlantic Ocean （78°S-50°N, 100°W-20°E） with HYCOM forced with atmospheric reanalysis from 1 January to 30 June 2010. Five integra- tions were performed, including the control run without assimilation. In the other four, different observations were assimilated： SST only （A SST）; Argo T-S profiles only （AArgo）; along-track SLA only （A_SLA）; and all data employed in the previous runs （A_All）. The A_SST, A_Argo, and A_SLA runs were very effective in improv- ing the representation of the assimilated variables, but they had relatively little impact on the variables that were not assimilated. In particular, only the assimilation of S was able to reduce the deviation of S with respect to ob- servations. Overall, the A_All run produced a good analy- sis by reducing the deviation of SST, T, and S with respect to the control run by 39%, 18%, and 30%, respectively, and by increasing the correlation of SLA by 81%.

Two-Hop Gaussian Relay Channel with Linear Relaying： Achievable Rate and Optimization Design

The relay node with linear relaying transmits the linear combination of its past received signals.The optimization of two-hop relay channel with linear relaying is discussed in this paper.The capacity for the two-hop Gaussian relay channel with linear relaying is derived,which can be formulated as an optimization problem over the relaying matrix and the covariance matrix of the signals transmitted at the source.It is proved that the solution to this optimization problem is equivalent to a "single-letter" optimization problem.We also show that the solution to this "single-letter" optimization problem has the same form as the expression of the rate achieved by Time-Sharing Amplify and Forward(TSAF).In order to solve this equivalent problem,we proposed an iterative algorithm.Simulation results show that if channel gain of one hop is relatively smaller,the achievable rate with TSAF is closer to the max-flow min-cut capacity bound,but at a lower complexity.

Study on Complete Analysis of LRE Test Samples Based on PCA

Incomplete data samples have a serious impact on the effectiveness of data mining.Aiming at the LRE historical test samples,based on correlation analysis of condition parameter,this paper introduced principle component analysis(PCA)and proposed a complete analysis method based on PCA for incomplete samples.At first,the covariance matrix of complete data set was calculated;Then,according to corresponding eigenvalues which were in descending,a principle matrix composed of eigen-vectors of covariance matrix was made;Finally,the vacant data was estimated based on the principle matrix and the known data.Compared with traditional method validated the method proposed in this paper has a better effect on complete test samples.An application example shows that the method suggested in this paper can update the value in use of historical test data.

Multi-sensor federated unscented Kalman filtering algorithm in intermittent observations

Aiming at the adverse effect caused by limited detecting probability of sensors on filtering preci- sion of a nonlinear system state, a novel muhi-sensor federated unscented Kalman filtering algorithm is proposed. Firstly, combined with the residual detection strategy, effective observations are cor- rectly identified. Secondly, according to the missing characteristic of observations and the structural feature of unscented Kalman filter, the iterative process of the single-sensor unscented Kalman filter in intermittent observations is given. The key idea is that the state estimation and its error covariance matrix are replaced by the state one-step prediction and its error covariance matrix, when the phe- nomenon of observations missing occurs. Finally, based on the realization mechanism of federated filter, a new fusion framework of state estimation from each local node is designed. And the filtering precision of system state is improved further by the effective management of observations missing and the rational utilization of redundancy and complementary information among multi-sensor observa- tions. The theory analysis and simulation results show the feasibility and effectiveness of the pro- posed algorithm.

Measurement of Incompatible Probability in Information Retrieval:A Case Study with User Clicks

The incompatible probability represents an important non-classical phenomenon, and it describes conflicting observed marginal probabilities, which cannot be satisfied with a joint probability. First, the incompatibility of random variables was defined and discussed via the non-positive semi-definiteness of their covariance matrixes. Then, a method was proposed to verify the existence of incompatible probability for variables. A hypothesis testing was also applied to reexamine the likelihood of the observed marginal probabilities being integrated into a joint probability space, thus showing the statistical significance of incompatible probability cases. A case study with user click-through data provided the initial evidence of the incompatible probability in information retrieval (IR), particularly in user interaction. The experiments indicate that both incompatible and compatible cases can be found in IR data, and informational queries are more likely to be compatible than navigational queries. The results inspire new theoretical perspectives of modeling the complex interactions and phenomena in IR.

Principal Component Analysis （PCA） on Multivariate Data of Lard Analysis in Cooking Oil

Discrimination of fatty acids （FAs） of lard in used cooking oil is important in halal determination. The aim of this study was to find the information related to the changes FAs of lard when frying in cooking oil. Quantitative analysis of FAs composition extracted from a series of experiments which involving frying cooking oil spiked with lard at three different parameters; concentration of spiked lard, heating temperatures and period of frying. The samples were analyzed using Gas Chromatography （GC） and Principal Components Analysis （PCA） technique. Multivariate data from chromatograms of FAs were standardized and computed using Unscrambler X10 into covariance matrix and eigenvectors correspond to Principal Components （PCs）. Results have shown that the first and second PCs contribute to the FAs mapping which can be visualized by scores and loading plots to discriminate FAs of lard in used cooking oil

Two Stage Estimation and Its Covariance Matrix in Multivariate Seemingly Unrelated Regression System

Multivariate seemingly unrelated regression system is raised first and the two stage estimation and its covariance matrix are given. The results of the literatures[1-5] are extended in this paper.

Strong Convergence of Empirical Distribution for a Class of Random Matrices

Let {vij}, i, j = 1, 2, …, be i.i.d, random variables with Ev11 = 0, Ev11^2 = 1 and a1 = （ai1,…, aiM） be random vectors with {aij} being i.i.d, random variables. Define XN =（x1,…, xk） and SN =XNXN^T,where xi=ai×si and si=1/√N（v1i,…, vN,i）^T. The spectral distribution of SN is proven to converge, with probability one, to a nonrandom distribution function under mild conditions.

Maximum Likelihood Blind Separation of Convolutively Mixed Discrete Sources

In this paper,a Maximum Likelihood(ML) approach,implemented by Expectation-Maximization(EM) algorithm,is proposed to blind separation of convolutively mixed discrete sources.In order to carry out the expectation procedure of the EM algorithm with a less computational load,the algorithm named Iterative Maximum Likelihood algorithm(IML) is proposed to calculate the likelihood and recover the source signals.An important feature of the ML approach is that it has robust performance in noise environments by treating the covariance matrix of the additive Gaussian noise as a parameter.Another striking feature of the ML approach is that it is possible to separate more sources than sensors by exploiting the finite alphabet property of the sources.Simulation results show that the proposed ML approach works well either in determined mixtures or underdetermined mixtures.Furthermore,the performance of the proposed ML algorithm is close to the performance with perfect knowledge of the channel filters.

An infrared human face recognition method based on 2DPCA

Aimed at the problems of infrared image recognition under varying illumination,face disguise,etc.,we bring out an infrared human face recognition algorithm based on 2DPCA.The proposed algorithm can work out the covariance matrix of the training sample easily and directly;at the same time,it costs less time to work out the eigenvector.Relevant experiments are carried out,and the result indicates that compared with the traditional recognition algorithm,the proposed recognition method is swift and has a good adaptability to the changes of human face posture.