基于小波和矩的图像字符特征提取方法研究

提取稳定而有代表性的特征是视频图像字符识别的核心问题之一。文章提出了一种基于小波和矩的图像字符特征向量提取方法。通过对字符图像的不同小波分解子图求取不同的矩特征,构造出字符的特征向量。该方法将小波对图像结构精细特征的把握能力强的优点与矩所具有的平移,缩放和旋转不变及抗噪性强的特性有机地结合起来,特征向量稳定、识别准确率高、算法快、抗噪性能强,且特征提取方法具有类人视觉特点。

基于Hu不变矩和BP网络的条形码图像识别方法

针对目前比较流行的一维条形码和二维条形码识别算法存在对几何失真图像的识别准确率较低的问题,提出了一种新的基于不变矩和BP网络的条形码识别方法,提取不变矩特征向量作为特征值输入BP网络,对其进行训练与测试,利用训练好的BP网络对形变条形码图像进行识别,实现了对存在旋转、平移和缩放等几何失真的条形码图像的正确识别。实验结果表明,基于Hu不变矩和BP网络的条形码识别方法具有很强的抗图像平移、拉伸和旋转识别能力,并且具有实现简单、训练速度快、识别率高等特点。

基于稀疏自编码器的混合信号符号检测研究

基于深度神经网络(DNN)的符号检测器(SD)的结构直接影响检测精度和计算复杂度,然而,已有的工作中并未对DNN符号检测器的结构选择方法开展研究。此外,已知的基于DNN的符号检测器复杂度较高且仅能完成单一调制信号的检测。针对以上问题,该文提出基于误符号率(SER)度量的低复杂度稀疏自编码器符号检测器(SAED)结构选择策略,同时,利用提出的累积量和矩特征向量(CMFV)实现了对混合信号的检测。所设计的符号检测器不依赖信道模型和噪声假设,对不同调制方式的信号具有较好的检测性能。仿真结果表明,该文设计的SAE符号检测器的SER性能接近最大似然(ML)检测理论值,且在频偏、相偏和有限训练样本等非理想条件下具有较强的鲁棒性。

Interface-Optical-Phonon Modes in Quasi-one-dimensional Wurtzite Rectangular Quantum Wires

By employing the dielectric continuum model and Loudon＇s uniaxial crystal model, the interface optical （IO） phonon modes in a freestanding quasi-one-dimensional （Q1D） wurtzite rectangular quantum wire are derived and analyzed. Numerical calculation on a freestanding wurtzite GaN quantum wire is performed. The resulte reveal that the dispersion frequencies of IO modes sensitively depend on the geometric structures of the Q1D wurtzite rectangular quantum wires, the free wave-number kz in z-direction and the dielectric constant of the nonpolar matrix. The degenerating behavior of the IO modes in Q1D wurtzite rectangular quantum wire has been clearly observed in the case of small wave-number kz and Iarge ratio of length to width of the rectangular crossing profile. The limited frequency behaviors of IO modes have been analyzed deeply, and detailed comparisons with those in wurtzite planar quantum wells and cylindrical quantum wires are also done. The present theories can be looked on as a generalization of that in isotropic rectangular quantum wires, and it can naturally reduce to the case of Q1D isotropic quantum wires once the anisotropy of the wurtzite material is ignored.

Novel histogram descriptor for global feature extraction and description

A novel histogram descriptor for global feature extraction and description was presented. Three elementary primitives for a 2×2 pixel grid were defined. The complex primitives were computed by matrix transforms. These primitives and equivalence class were used for an image to compute the feature image that consisted of three elementary primitives. Histogram was used for the transformed image to extract and describe the features. Furthermore, comparisons were made among the novel histogram descriptor, the gray histogram and the edge histogram with regard to feature vector dimension and retrieval performance. The experimental results show that the novel histogram can not only reduce the effect of noise and illumination change, but also compute the feature vector of lower dimension. Furthermore, the system using the novel histogram has better retrieval performance.

Inverse Eigenvalue Problems for a Structure with Linear Parameters

The inverse design method of a dynamic system with linear parameters has been studied. For some specified eigenvalues and eigenvectors, the design parameter vector which is often composed of whole or part of coefficients of spring and mass of the system can be obtained and the rigidity and mass matrices of an initially designed structure can be reconstructed through solving linear algebra equations. By using implicit function theorem, the conditions of existence and uniqueness of the solution are also deduced. The theory and method can be used for inverse vibration design of complex structure system.

Principal component analysis using neural network

The authors present their analysis of the differential equation d X(t)/ d t = AX(t)-X T (t)BX(t)X(t) , where A is an unsymmetrical real matrix, B is a positive definite symmetric real matrix, X ∈R n; showing that the equation characterizes a class of continuous type full feedback artificial neural network; We give the analytic expression of the solution; discuss its asymptotic behavior; and finally present the result showing that, in almost all cases, one and only one of following cases is true. 1. For any initial value X 0∈R n, the solution approximates asymptotically to zero vector. In this case, the real part of each eigenvalue of A is non positive. 2. For any initial value X 0 outside a proper subspace of R n, the solution approximates asymptotically to a nontrivial constant vector (X 0) . In this case, the eigenvalue of A with maximal real part is the positive number λ=‖(X 0)‖ 2 B and (X 0) is the corresponding eigenvector. 3. For any initial value X 0 outside a proper subspace of R n, the solution approximates asymptotically to a non constant periodic function (X 0,t) . Then the eigenvalues of A with maximal real part is a pair of conjugate complex numbers which can be computed.

Action Recognition from a Different View

In this paper,we propose a novel approach to recognise human activities from a different view.Although appearance-based recognition methods have been shown to be unsuitable for action recognition for varying views,there must be some regularity among the same action sequences of different views.Selfsimilarity matrices appear to be relative stable across views.However,the ability to effectively realise this stability is a problem.In this paper,we extract the shape-flow descriptor as the low-level feature and then choose the same number of key frames from the action sequences.Self-similarity matrices are obtained by computing the similarity between any pair of the key frames.The diagonal features of the similarity matrices are extracted as the highlevel feature representation of the action sequence and Support Vector Machines(SVM) is employed for classification.We test our approach on the IXMAS multi-view data set.The proposed approach is simple but effective when compared with other algorithms.

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.

Some Remarks for Discrete Versions of Nodal Domain Theorems

In this paper, an error is firstly pointed out in the proof of the main theorems (Theorem 4 and Theorem 6) in [1]. Then the error is corrected and the right proof is given.

PROJECTION-PURSUIT BASED PRINCIPAL COMPONENT ANALYSIS:A LARGE SAMPLE THEORY

The principal component analysis （PCA） is one of the most celebrated methods in analysing multivariate data. An effort of extending PCA is projection pursuit （PP）, a more general class of dimension-reduction techniques. However, the application of this extended procedure is often hampered by its complexity in computation and by lack of some appropriate theory. In this paper, by use of the empirical processes we established a large sample theory for the robust PP estimators of the principal components and dispersion matrix.

ON ASYMPTOTIC JOINT DISTRIBUTIONS OF EIGENVALUES OF RANDOM MATRICES WHICH ARISE FROM COMPONENTS OF COVARIANCE MODEL

In this paper, the authors derive the asymptotic joint distributions of theeigenvalues of some random matrices which arise from components of covariance model.