GRU

不错的svm讲义，对于svm的背景和原理有比较详细的讲解(svm)

使用中科院虹膜数据库，实现了虹膜内外边界定位(Iris database using the Chinese Academy of Sciences, to achieve both inside and outside the iris boundary localization)

用matlab产生粉红色噪声和高斯色噪声,对粉红色噪声和高斯色噪声进行相关分析和谱分析,所有结果均用图示法来表示.(Using matlab to generate pink noise and Gaussian colored noise, pink noise and Gaussian colored noise correlation analysis and spectrum analysis, all the results were that uses graphics to represent.)

说明：  模拟退火工具箱，程序主体和子函数含大量注解内容，非常适合学习者了解该算法的结构。 (Simulated Annealing toolbox, procedures and Functions with the main contents of a large number of comments, very suitable for learners to understand the structure of the algorithm.)

參考文獻「Gabor Feature based Sparse Representation for Face Recognition with Gabor Occlusion Dictionary」and matlab code.(「Gabor Feature based Sparse Representation for Face Recognition with Gabor Occlusion Dictionary」and matlab code.)

在Visual+C中调用Matlab的其他方法.pdf(In the Visual+ C to call the other methods Matlab. Pdf)

杜达的《模式分类》第二版配套的matlab代码，供做该方面研究的同学学习！(the matlab code of 《 pattern classification》)

ABC_FDTD_Die(T) Implements simulation of a Gaussian Pulse over T time steps. ABC are for free space. If boundaries are in the Dielectric medium then the ABC fail. Dielectric medium begin and end can be specified with the code(ABC_FDTD_Die (T) Implements simulation of a Gaussian Pulse over T time steps. ABC are for free space. If boundaries are in the Dielectric medium then the ABC fail. Dielectric medium begin and end can be specified with the code)

dIVISIVE CLUSTER ALGORITHM

鉴于LCD方法存在的问题，本文提出了一种基于互相关匹配端点延拓局部特征尺度分解（Cross-correlation matching endpoint Extension Local Characteristic scale Decomposition，简称CELCD），由于LCD分解原理是依据信号的局部极值点信息不断进行筛分信号，在信号分解时需要先确定信号的局部极值点，而信号的两个端点可能不是极值点，因此在信号两端点就会出现虚假成分，且该现象随着分解的进行向数据内部扩散，产生端点效应，导致分解结果失真。而CELCD方法的本质在于探索待分解信号前后端的数据规律，延长波形，使得端点效应不出现在原始信号分解的片段上，从而消除端点效应。(Given the LCD method has a problem, this paper presents a cross-correlation based on local feature matching endpoint extension scale decomposition (Cross-correlation matching endpoint Extension Local Characteristic scale Decomposition, referred CELCD), since the LCD decomposition principle is based on the signal of local optima ongoing screening point information signal, when the signal decomposition needed to determine local extreme point of the signal, and the two endpoints signals may not be extreme point, so the two end points of the signal will be fraudulent, and that the phenomenon with decomposition of the data in the internal diffusion, resulting end effect, leading to decomposition distortion. The essence of the method is to explore CELCD data signal to be decomposed before and after the end of the law, to extend the waveform, so the end effect does not appear on the original signal decomposition fragments now to eliminate the end effect.)