FACE-Recognition

第8章 图象的检测及模板匹配 图象的分割与检测(识别)实际上是一项非常困难的工作。很难说清楚为什么图象应该分割成这样而不是那样。(Chapter 8 Detection of image template matching and image segmentation and detection (recognition) is actually a very difficult task. It is hard to say clearly why the image should be divided into such and not as.)

随着生产和工艺的进步，人们对产品的质量要求越来越高，基于机器视觉的在线 检测系统成为一种重要的质量控制手段。本文对应用于宽幅面、高精度的基于分布式 机器视觉的产品表面缺陷在线检测理论与算法进行了研究，并在此基础上以印刷品和 浮法玻璃的缺陷检测为例，完成了模式产品和非模式产品的缺陷在线检测系统。(With the progress of production and technology, people are increasingly demanding high quality products, based on machine vision for online 检测 system into one kind of Zhong Yao quality Kongzhishouduan. This article applies to wide format, high-precision machine vision products based on distributed-line surface defect detection theory and algorithms are studied, and on this basis, and float glass in print defect detection, for example, completed a model of products and non-model product line defect detection system.)

基于SIFT+Kmeans+LDA的图片分类器的实现源码。 博文参考：http://www.cnblogs.com/freedomshe/archive/2012/04/24/2468747.html(The pictures classifier based on SIFT, Kmeans and LDA. Blog Reference: http://www.cnblogs.com/freedomshe/archive/2012/04/24/2468747.html)

) 使用分块的主成分分析方法(PCA)对人脸图像进行压缩编码。针对PCA方法计算量大的缺点,首先把问题转化成奇异值分解(SVD)问题,然后设计了特征空间的更新算法,通过递推,简化每一步计算的计算量,达到了实时编码的要求。 4) 在Windows平台下基于Video for Windows(VFW)接口开发了人脸视频图像编码和解码的实验系统,该系统实现了图像采集、图像显示、编码、解码等功能。() The use of sub-blocks of principal component analysis (PCA) on the human face image coding. PCA method for large defects, first of all the issues into a singular value decomposition (SVD) problem, and then design a feature space of the update algorithm, recursive, simplifying the calculation of the amount calculated at each step to achieve real-time encoding requirements. 4) In the Windows platform based on Video for Windows (VFW) interfaces, the human face image coding and decoding video experimental system, the system achieved image acquisition, image display, encoding, decoding functions.)

过特定人语音情感数据库的建立；语音情感特征提取；语音情感分类器的设计，完成了一个特定人语音情感识别的初步系统。对于单个特定人，可以识别平静、悲伤、愤怒、惊讶、高兴5种情感，除愤怒和高兴之间混淆程度相对较大之外，各类之间区分特性良好，平均分类正确率为93.7 。对于三个特定人组成的特定人群，可以识别平静、愤怒、悲伤3种情感，各类之间区分特性良好，平均分类正确率为94.4 。其中分类器采用混合高斯分布模型。(The system of speech emotion recognition)

一个集成的图形界面的程序，可调用每一次作业子程序。 一、调用画点的函数，用DDA、中点算法画直线和中点算法画圆和椭圆。 二、多边形扫描转换算法和区域填充算法实现（扫描线算法为必做，基于求余运算的边缘填充和边标志算法为任选；基于种子的区域填充采用4连通区域的递归种子填充算法，或扫描线种子填充算法，要求种子点(x,y)可交互输入）。 三、线段裁剪和多边形裁剪算法的动画演示实现。（两种线段裁剪算法和H-S多边形逐边裁剪算法）多边形裁剪算法的动画演示要求先画出一个封闭的多边形，再画矩形的裁剪窗口，然后选择裁剪按钮（或命令），按下“上边裁剪”按钮（或执行“上边裁剪”命令），多边形相对裁剪窗口的上边进行裁剪，显示上边裁剪后的多边形，依此进行其它各边裁剪。 四、用动画实现二维图形变换的各种算法，实现对指定形体的平移、旋转和缩放。（包括类似自行车行走和绕固定点旋转的自旋转物体动画。） 五、简单三维图形系统：凸多面体的建模、透视投影，隐藏面的消除及基本图形变换（平移、旋转、缩放）。 六、交互式Bezier曲线的输入绘制程序实现 (An integrated graphical interface of the program, each operation subroutine callable. One call a function of the point of the painting, with the DDA, the midpoint algorithm for drawing a straight line and the midpoint algorithm circle and oval. Second, the polygon scan conversion algorithm and region filling algorithm (scan line algorithm for must fill and edge flag algorithm based on the edge of the modulo operator is optional using the 4 regional connectivity recursive seed filling algorithm based on the area of seed filling, or scan line seed filling algorithm requires the seed point (x, y) can be entered interactively). Line Clipping polygon clipping algorithm animation achieved. (Two Line Clipping Algorithm and HS polygon-by-side clipping algorithm) polygon clipping algorithm animation requires draw a closed polygon, rectangle and then draw the crop window, and then select the the cropped button (or command), press the "on top of cutting "button (or the implementation of)

CCD 摄像机内外部参数快速准确的标定方法(Fast and accurate calibration method for internal and external parameters of CCD camera)

opencv c++将彩色图像转换为灰度图像进行边缘检测，(opencv c++ to convert color images to gray-scale image edge detection,)

输入一个带权无向图的顶点数、边数、各条边信息（两个顶点和权值），建立该图的邻接矩阵结构，输出该邻接矩阵。将上述无向图邻接矩阵转换为邻接表结构，输出该邻接表；根据该邻接表对无向图进行深度优先遍历序列和广度优先遍历序列，并输出遍历结果；用prim算法实现构造该带权无向图的最小生成树，并将该最小生成树的各条边信息输出。(Enter a weighted undirected graph with vertices, edges, each of the side information (two vertices and weights), the establishment of the adjacency matrix structure, the output of the adjacency matrix. The above undirected graph adjacency matrix into adjacency list structure, the output of the adjacency list according to the adjacency table undirected graph depth-first traversal sequence and breadth-first traversal sequence, and outputs through the results using prim algorithm construct the weighted undirected graph minimum spanning tree, and the minimum spanning tree information on respective sides of the output.)

图片处理 去噪音，去模糊为了实现较好保留图像边缘特征的有效去噪,提出了一种基于边缘检测的Contourlet 变换去噪方法(Picture Deal)