PLL_System_Sim_for_matlab

　计算机人脸识别技术( Face Reocgnition)就利用计算机分析人脸图像,从中提取出有效的识别信息,用来辨认身份的一门技术。[ 1 ]即对已知人脸进行标准化处理后,通过某种方法和数据库中的人脸样本进行匹配,寻找库中对应人脸及该人脸相关信息。人脸自动识别系统有两个主要技术环节,一是人脸定位,即从输入图像中找到人脸存在的位置,将人脸从背景中分割出来,二是对标准化后的人脸图像进行特征提取和识别。本文中介绍的PCA (特征脸)方法就是一种常用的人脸 特征提取方法。(Computer Face Recognition Technology (Face Reocgnition) on the use of computer analysis of facial image, to extract the valid identification information used to identify the status of a technology. [1] that is known to standardize treatment of face, through a method and a database of face samples for matching, search library, the corresponding face and the face-related information. Automatic face recognition system has two main technical aspects, first, face location, that is, from the input image to find the location of the face there, the faces will be split out from the background, the second is, the standard features of face images extraction and recognition. Described in this paper PCA (Eigenfaces) method is a common facial feature extraction method.)

Pyramid.zip包含 1、pyramid.fig界面 2、高斯金字塔通过GPReduce.m函数实现 拉普拉斯金字塔由GPExpand.m函数实现 3、用imshow()显示拉普拉斯金字塔时，如果图像矩阵数据类型为double，会显示为彩色；而通过uint8()将类型强制转换为8位无符号整数，图像显示就为黑白了。 4、采用256*256的lena彩色图像(Pyramid.zip coantain 1、pyramid.fig interface 2、implemented by GPReduce.m and GPExpand.m 3、use 256*256 lena color picture)

non linear optimization technic

牛顿插值法，可用于非线性拟合，并返回在某点出的近似函数值。(Newton interpolation method can be used for nonlinear fitting, and the function returns the approximate value at some point out.)

用三种不同的Runge-kutta方法计算常微分方程(Runge-kutta for ODE)

说明：  matlab代码，新进的emd分解工具，比小波效果好(Matlab code, the new EMD decomposition tool, is better than wavelet)

数值分析中的高斯消元法利用列主元进行消元程序(Numerical analysis using the Gaussian elimination method carried out principal component elimination process)

基于差分进化的粒子群优化算法，（全局最优粒子群优化算法）(Based on differential evolution particle swarm optimization, (global optimization PSO))

matlab完整源代码，可以实现车牌自动识别。(Matlab complete source code, can realize automatic license plate recognition.)

Road extraction by gradient density