Evaluation of a set of new ORF kernel functions of SVM for speech recognition

The kernel function is the core of the Support Vector Machine (SVM), and its selection directly affects the performance of SVM. There has been no theoretical basis on choosing a kernel function for speech recognition. In order to improve the learning ability and generalization ability of SVM for speech recognition, this paper presents the Optimal Relaxation Factor (ORF) kernel function, which is a set of new SVM kernel functions for speech recognition, and proves that the ORF function is a Mercer kernel function. The experiments show the ORF kernel function's effectiveness on mapping trend, bi-spiral, and speech recognition problems. The paper draws the conclusion that the ORF kernel function performs better than the Radial Basis Function (RBF), the Exponential Radial Basis Function (ERBF) and the Kernel with Moderate Decreasing (KMOD). Furthermore, the results of speech recognition with the ORF kernel function illustrate higher recognition accuracy.     

基于支持向量机的传真收件人识别方法

在字符特征提取基础上,文章提出了应用支持向量机对传真收件人进行识别的方案,解决了传真收件人格式、表示方法多样性而导致的自动分发困难的问题。文中对四种常用的核函数分别进行了实验,选取了对传真收件人具有较高识别率的核函数,它有利于实现传真文件的自动分发。     

基于SVM与非参数LDA 的雷达自动目标识别

针对非参数线性判别分析（LDA）的类间散布矩阵，就如何有效描述类边界结构这一问题，提出一种SVM与k近邻（kNN）法相结合的非参数类间散布矩阵构造方法——SVM—kNN．该方法消除了非类边界样本对类边界结构信息的扭曲．将SVM—kNN非参数LDA方法用于外场实测高分辨距离像的特征提取，并将识别结果与加权kNN非参数LDA法和谱域原空间法比较，结果表明，SVM—kNN非参数LDA方法能显著提高识别效率．     

多类分类预选取的SVM在语音识别中的应用

支持向量机在训练过程中,将很多时间都浪费在对非支持向量的复杂计算上,特别是对于大规模数据量的语音识别系统来说,支持向量机在训练时间上不必要的开销将会更加显著。核模糊C均值聚类是一种常用的典型动态聚类算法,并且有核函数能够把模式空间的数据非线性映射到高维特征空间。在核模糊C均值聚类的基础上,结合了多类分类支持向量机中的一对一方法,按照既定的准则把训练样本集中有可能属于支持向量的样本数据进行预选取,并应用到语音识别中。实验取得了较好的结果,该方法有效地提高了支持向量机分类器的学习效率和泛化能力。     

基于新型特征提取的寄生虫卵图像识别研究

讨论了用支持向量机进行多分类的若干学习策略,提出了一种新型图像特征提取方法,以此来实现对鞭虫等九种寄生虫卵图像自动识别和分类,平均识别率优于传统神经网络,达到了93.9%,为寄生虫卵图像识别提供了一种新方法。     

基于ICA-MJE和SVM的虹膜特征提取与识别

提出了一种新的虹膜特征提取与识别方法。对虹膜纹理采用最大判别熵的独立分量分析(ICA-MJE)实现特征提取,通过支持向量机(SVM)完成模式匹配。与Gabor小波的方法比较，在编码长度和编码时间方面有明显地改进。实验结果表明，该算法能更好地提高虹膜的识别率并能够有效地应用于身份识别系统中。     

EMD在语音情感识别中的应用与研究

语音情感计算引起了国内外广泛的关注,特别是在语音情感特征提取方面做了大量的研究。利用经验模态分解(EMD)方法对情感语音进行处理,得到情感语音的前4阶固有模态函数(IMF),并将前4阶IMF分别通过Hilbert变换得到其瞬时频率和瞬时振幅。提取它们的统计特征,再结合情感语音的声学特征共同组成情感特征向量,并对特征向量做归一化处理。利用支持向量机(SVM)对四种情感语音即生气、高兴、悲伤和平静进行识别。实验结果表明该方法的识别效果较好。     

基于语音识别的汉语发音自动评分系统的设计与实现

语音识别技术的发展使得人与计算机的交互成为可能,针对目前对外汉语中发音教学的不足,在结合了语音识别的相关原理,提出了在对外汉语教学领域中汉语自动发音水平评价系统的设计,详细地描述了系统的结构、功能及流程.介绍了系统实现中的关键技术和步骤:动态时间弯折算法、语料库的建立、声韵分割技术以及评价分级标准.通过小范围的试验,表明该系统对留学生汉语发音水平的测试有一定的参考价值.     

基于智能语音的翻译机器人自动化控制系统设计

为提升自动控制效果,加快翻译速率,设计基于智能语音的翻译机器人自动化控制系统。采集外界智能语音信号,利用A/D转换器得到数字信号,启动语音唤醒模块激活翻译机器人,听写模式识别复杂语音信号,命令模式识别简单语音信号,得到语言文本识别结果,通过深度学习关键词检测方法提取关键词作为翻译机器人的自动化控制指令,通过单片机识别自动化控制指令。实验结果表明,该系统可有效采集外界智能语音信号,提取智能语音信号的关键词,完成翻译机器人自动化控制。     

Development of a personified face emotion recognition technique using fitness function

In this article, two subjects, one South East Asian (SEA) and the other Japanese, are considered for face emotion recognition using a genetic algorithm (GA). The parameters relating the face emotions in each case are entirely different. All six universally accepted emotions and one neutral are considered for each subject. Eyes and lips are usually considered as the features for recognizing emotions. This paper has two parts. The first part investigates a set of image processing methods suitable for recognizing face emotion. The acquired images have gone through a few preprocessing methods such as gray-scale, histogram equalization, and filtering. The edge detection has to be sufficiently successful even when the light intensity is uneven. So, to overcome this problem, the histogram-equalized image has been split into two regions of interest (ROI): the eye and lip regions. The two regions have been applied with the same preprocessing methods, but with different threshold values. It was found that the Sobel edge detection method performed very well in segmenting the image. Three feature extraction methods are considered, and their respective performances are compared. The method which is fastest in extracting eye features is adopted. The second part of the paper discusses the way to recognize emotions from eye features alone. Observation of various emotions of the two subjects lead to an unique eye characteristic, that is, the eye exhibits ellipses of different parameters in each emotion. The GA is adopted to optimize the ellipse characteristics of the eye features in each emotion based on an ellipse-based fitness function. This has shown successful emotion classifications, and a comparison is made on the emotions of each subject.     

Image processing based automatic recognition of asphalt pavement patch using a metaheuristic optimized machine learning approach

Patch detection is an important task in pavement condition survey. This study establishes an automatic approach for asphalt pavement patch recognition based on image texture analysis and hybrid machine learning algorithms. Features based on image texture that employs statistical properties of color channels and the gray-scale co-occurrence matrix are used by the Least Squares Support Vector Machine (LSSVM) for discriminating patched areas from non-patch ones. In addition, to optimize the LSSVM training phase, the Differential Flower Pollination (DFP) metaheuristic is used. A data set constructed from a set of 1000 image samples has been utilized to train and verify the proposed integration of image texture analysis techniques, LSSVM, and DFP. Experimental results show that the new model can achieve a good prediction result with Classification Accuracy Rate = 95.30%, Positive Predictive Value = 0.96, and the Negative Predictive Value = 0.95. Additionally, a patch detection program has been developed and compiled in Visual C# .NET to ease the implementation of the hybrid model. Thus, the newly developed method can be a potential tool for traffic management agencies during the phase of pavement condition evaluation.     

Feature extraction of three-dimensional objects and visual processing in a hand-eye system using laser tracker

Visual processing is one of the important aspects in researches on an intelligent robot. This paper describes the principle and the construction of a laser tracker as a visual device with an active faculty. This robot's eye can acquire three-dimensional coordinates of a laser spot on an object based on triangulation and then extract features of the object corresponding to the purpose of the intelligent rogot. As its applications, classification of several three-dimensional objects and visual processing in a hand-eye system are performed.