Local refinement of three-dimensional finite element meshes

Mesh refinement is an important tool for editing finite element meshes in order to increase the accuracy of the solution. Refinement is performed in an iterative procedure in which a solution is found, error estimates are calculated, and elements in regions of high error are refined. This process is repeated until the desired accuracy is obtained.Much research has been done on mesh refinement. Research has been focused on two-dimensional meshes and three-dimensional tetrahedral meshes ([1] Ning et al. (1993) Finite Elements in Analysis and Design, 13, 299–318; [2] Rivara, M. (1991) Journal of Computational and Applied Mathematics 36, 79–89; [3] Kallinderis; Vijayar (1993) AIAA Journal,31, 8, 1440–1447; [4] Finite Element Meshes in Analysis and Design,20, 47–70). Some research has been done on three-dimensional hexahedral meshes ([5] Schneiders; Debye (1995) Proceedings IMA Workshop on Modelling, Mesh Generation and Adaptive Numerical Methods for Partial Differential Equations). However, little if any research has been conducted on a refinement algorithm that is general enough to be used with a mesh composed of any three-dimensional element (hexahedra, wedges, pyramids, and/or retrahedra) or any combination of three-dimensional elements (for example, a mesh composed of part hexahedra and part wedges). This paper presents an algorithm for refinement of three-dimensional finite element meshes that is general enough to refine a mesh composed of any combination of the standard three-dimensional element types.     

Automatic gender recognition based on pixel-pattern-based texture feature

A pixel-pattern-based texture feature (PPBTF) is proposed for real-time gender recognition. A gray-scale image is transformed into a pattern map where edges and lines are to be used for characterizing the texture information. On the basis of the pattern map, a feature vector is comprised the numbers of the pixels belonging to each pattern. We use the image basis functions obtained by principal component analysis (PCA) as the templates for pattern matching. The characteristics of the feature are comprehensively analyzed through an application to gender recognition. Adaboost is used to select the most discriminative feature subset, and support vector machine (SVMs) is adopted for classification. Performed on frontal images from FERET database, the comparisons with Gabor show that PPBTF is a significant facial representation, quite effective and speedier in computation.     

基于三维卷积神经网络的虫音特征识别方法

进口木材蛀虫检疫是海关的一项重要工作,但其存在着虫声检测算法准确率低、鲁棒性差等问题。针对这些问题,提出了一种基于三维卷积神经网络（3D CNN）的虫音检测方法以实现虫音特征的识别。首先,对原始虫音音频进行交叠分帧预处理,并使用短时傅里叶变换得到虫音音频的语谱图;然后,将语谱图作为3D CNN的输入,使其通过包含三层卷积层的3D CNN以判断音频中是否存在虫音特征。通过设置不同分帧长度下的输入进行网络训练及测试;最后以准确率、F₁分数以及ROC曲线作为评估指标进行性能分析。结果表明,在交叠分帧长度取5 s时,训练及测试效果最佳。此时,3D CNN模型在测试集上的准确率达到96.0%,F₁分数为0.96,且比二维卷积神经网络（2D CNN）模型准确率提高近18%。说明所提算法能准确地从音频信号中提取虫音特征并完成蛀虫识别任务,为海关检验检疫提供有力保障。     

基于改进的NC-HOG特征的工程车车型自动识别算法

为有效识别工程车车型,提出一种改进的HOG特征的自动识别算法.使用RPN(region proposal network)自动划分车辆候选区,对每个候选区进行自然度保留的图像增强处理和颜色不变性处理,分别提取NPE-HOG(naturalness preserved enhancement-histogram of oriented gradients)特征和CIV-HOG(color invariant-histogram of oriented gradients)特征,将两者融合得到NC-HOG特征,结合一对一支持向量机实现对压路机、挖掘机、装载机3类工程车辆的车型自动识别.实验对国家电网施工现场的工程车数据库进行测试,对比从原图、Gamma校正、自然度保留图像增强和颜色不变性处理图像上分别提取HOG、LBP(local binary pattern)、SIFT(scale-invariant feature transform)特征对车型识别效果的影响.实验结果表明,改进NC-HOG算法能有效提高HOG特征的识别正确率,3种车型平均识别正确率为93.0％.     

基于特征面的分层特征识别

为降低特征识别的复杂度,提出基于特征实体、特征实面和特征虚面概念的层次性特征分类方法.通过构造2类神经网络输入矩阵,利用神经网络在特征识别中所具有的优势,实现基于特征面的分层特征识别方法.实例表明：该方法在识别去除材料的特征时比较有效,但识别特征的范围受到一定限制.     

A three-dimensional numerical manifold method based on tetrahedral meshes

A three-dimensional NMM based on tetrahedral meshes is developed in this paper. First, the displacement functions of manifold elements are formulated by the partition of unity function. Then, the global equilibrium equations for three-dimensional elasto-statics problem are established by minimizing the total potential energy. The stiffness matrix, loading matrix, displacement resistance matrix are derived for program coding. Thereafter, the computational cost of the global stiffness matrix in the NMM is discussed and compared with that in the classical finite element techniques. Finally, two typical examples are investigated to demonstrate the validity of the proposed method. The results indicate that an enough accuracy of approximation will be obtained with decreasing size of the mathematical mesh. The proposed method can be used as potentially powerful tool for three-dimensional structural analysis.     

Automatic a priori round-off analysis

The effect of rounding errors on an algebraic process is often investigated by means of a so-called backward analysis. In this paper we will discuss the possibility of performing such an analysis on a computer. We begin with a precise definition of a stable algorithm, i.e., an algorithm which is relatively insensitive to rounding errors.     

基于分形特征的植物识别

为了获得实用、可靠的植物识别方法,在识别植物识别过程中引入分形方法与分形特征.植物生长的过程在本质上与分形过程一致,都是在原有基础上进行迭代,这种迭代也可以推广到多株植物的情况.因此,分形特征不仅能区分单株植物的类别,也可以识别植物群落.通过多尺度求解图像分形特征,可以进一步检测杂株的分布.实验结果表明,基于分形特征的植物识别方法的识别率可以达到80%以上.     

Facial-expression recognition based on a low-dimensional temporal feature space

Multimedia Tools and Applications - This paper suggests a facial-expression recognition in accordance with face video sequences based on a newly low-dimensional feature space proposed. Indeed, we...     

Automatic feature generation for handwritten digit recognition

An automatic feature generation method for handwritten digit recognition is described. Two different evaluation measures, orthogonality and information, are used to guide the search for features. The features are used in a backpropagation trained neural network. Classification rates compare favorably with results published in a survey of high-performance handwritten digit recognition systems. This classifier is combined with several other high performance classifiers. Recognition rates of around 98% are obtained using two classifiers on a test set with 1000 digits per class     

Automatic real-time road marking recognition using a feature driven approach

Automatic road marking recognition is a key problem within the domain of automotive vision that lends support to both autonomous urban driving and augmented driver assistance such as situationally aware navigation systems. Here we propose an approach to this problem based on the extraction of robust road marking features via a novel pipeline of inverse perspective mapping and multi-level binarisation. A trained classifier combined with additional rule-based post-processing then facilitates the real-time delivery of road marking information as required. The approach is shown to operate successfully over a range of lighting, weather and road surface conditions.     

Emotional speech feature normalization and recognition based on speaker-sensitive feature clustering

In this paper we propose a feature normalization method for speaker-independent speech emotion recognition. The performance of a speech emotion classifier largely depends on the training data, and a large number of unknown speakers may cause a great challenge. To address this problem, first, we extract and analyse 481 basic acoustic features. Second, we use principal component analysis and linear discriminant analysis jointly to construct the speaker-sensitive feature space. Third, we classify the emotional utterances into pseudo-speaker groups in the speaker-sensitive feature space by using fuzzy k-means clustering. Finally, we normalize the original basic acoustic features of each utterance based on its group information. To verify our normalization algorithm, we adopt a Gaussian mixture model based classifier for recognition test. The experimental results show that our normalization algorithm is effective on our locally collected database, as well as on the eNTERFACE’05 Audio-Visual Emotion Database. The emotional features achieved using our method are robust to the speaker change, and an improved recognition rate is observed.     

基于显著特征和全局特征融合的植物识别方法

针对自然界中植物数据规模大且分布不平衡导致的识别困难的问题,提出一种基于显著特征和全局特征融合的植物识别方法.通过多层特征融合方法改进VGG19网络以提取植物的全局特征,并对全局特征进行显著特征提取得到图像的显著特征,将显著特征和全局特征进行自适应加权特征融合得到融合特征,对融合特征进行分类识别.在PlantCLEF2...     

新的基于特征关系表述的步态识别算法

为了快速有效地进行步态识别,利用特征关系非平稳分布的统计特性,提出了一种新的基于特征关系表述的步态识别算法。首先,将剪影轮廓相邻像素点间8邻域相对方向标号作为特征关系属性一,将轮廓边界点与中心点间的距离作为特征关系属性二,经直方图归一化处理,得到两种关系属性的联合概率;其次,结合主成分分析(PCA)降维的方法,提取特征主向量;最后,采用最近邻分类器进行识别分类。实验证明,该算法在CASIA步态数据库上,最高达到了90%以上的识别率,而且与传统的特征关系表述步态识别算法相比,关系属性联合概率矩阵维数由900维下降到240维,大大降低了算法的计算代价。     

基于时序约束建模的自动精化和组合工具

为提供更优质的使用Event-B形式化方法建模混合系统的工具,根据混合系统的时序约束建模方法,其能够很好刻画混合系统建模中的时间相关性质并且支持精化和组合,提出基于它的自动筛选、精化和组合的方法.开发对应的自动精化和组合的工具链,工具链包含模型检测、自动精化和组合、模型证明等一系列功能并拥有用户友好的界面.给出一个使用...     

基于词频同现与WordNet的图像自动标注改善算法研究

图像自动标注是模式识别与计算机视觉等领域中的重要问题。针对现有图像自动标注模型普遍受到语义鸿沟问题的影响,提出了基于关键词同现的图像自动标注改善方法,该方法利用数据集中标注词间的关联性来改善图像自动标注的结果。此外,针对上述方法不能反映更广义的人的知识以及易受数据库规模影响等问题,提出了基于语义相似的图像自动标注改善方法,通过引入具有大量词汇、包含了人知识的结构化电子词典WordNet来计算词汇间的关系并改善图像自动标注结果。实验结果表明,提出的两个图像自动标注改善方法在各项评价指标上相比以往模型均有所提高。     

基于自适应特征选取的步态识别

提出基于自适应特征选取的步态识别方法。采用新的特征提取方法,该方法将目标按人体结构特点划分为多个子区域,利用各个子区域的质心与头部质心的距离和夹角对步态特征进行描述。采用Boosting算法自适应选取最优特征序列,对识别结果进行加权处理。该方法结合了步态的动态和静态信息,实验结果表明该方法具有较高的识别性能。     

基于自适应特征融合的人脸识别

针对单一人脸特征在人脸识别中的局限性问题和二维主成分分析人脸特征缺少判别信息的问题,利用互补思想,提出了一种改进的二维主成分分析与二维线性鉴别分析加权融合的人脸识别算法。利用离散余弦变换对原始人脸图像进行压缩并重建,以滤除图像中人眼并不敏感的中高频部分,再利用二维主成分分析方法进行人脸特征的提取;运用二维线性鉴别分析方法提取原始人脸图像中具有鉴别性的人脸特征;最后,提出一种自适应的权值选取方法,将两种人脸特征进行加权融合以实现分类识别。在ORL和Yale人脸数据库上的实验结果证明了该方法的有效性。     

基于游程特征的文本图像识别方法

提出了一种基于游程特征的中英文本图像识别方法,用游程统计特征提供的图像信息作为图像模式识别的描述特征,在此基础上利用神经网络来训练分类器。实验结果表明,该方法的识别精度较高,具有一定的容错能力。     

基于特征点匹配的步态识别

基于图像特征点匹配的算法思想,结合步态能量图(GEI),提出了一种适用于2幅GEI匹配的步态识别方法.在GEI中采用改进的FAST算法提取特征点,并采用具有良好特征描述性能的BRIEF算法描述特征点.考虑到GEI匹配不要求特征点具有旋转不变性,提出了一种质心角约束条件加速特征点的匹配.在CASIA数据库B库上的实验结果表明:方法在识别率和特征计算时间上均具有良好的表现.