基于DM642的人脸识别系统毕业设计论文(原创).doc

毕业设计 学 院 计算机与信息学院 2012 年 6 月 12日基于DM642的人脸识别系统 摘 要 随着计算机处理速度的飞速提高和图像处理算法的革命性改进，作为生物特征识别技术的一个重要的研究方向---人脸识别技术在各种生物特征识别技术中脱颖而出，以其独特的主动性，非侵犯性和用户友好等众多优势受到世人瞩目。 本毕业设计是基于TI公司生产的DM642开放板设计的，在DM642这个硬件平台上实现了一个经典的基于PCA算法的简单的人脸识别系统，实现了简单的人脸识别。主要编程实现了以下几个功能：视频采集功能，人脸的检测和跟踪，图像去噪声，PCA算法的实现，以及人脸的识别分类。 本系统的设计思想：整个视频处理的过程是围绕一个视频采集框和一个人脸跟踪框设计的。只有当人脸跟踪框位于视频采集框内部的时候，才采集一帧图像到内存中。人脸识别过程主要分为两个阶段：训练阶段和识别阶段。 训练阶段，系统会采集大量的人脸图像进行训练，得到特征脸空间。然后将“人脸”投影到特征脸空间中。 识别阶段，系统会将采集到的待识别的人脸，也投影到特征空间中，得到该“人脸”在特征脸空间中的坐标值。然后对比各个坐标值的欧式距离，来达到识别的目的。 关键字：DM642，人脸识别, PCA，视频采集框，人脸跟踪框 A face recognition system based on DM642 ABSTRACT The prodigious development of computer speed and the revolutionary improvement of digital image processing is very developed. So the face recognition technology distinguish itself from other biometrical identification technologies for its active ,non-aggressive and user-friendly characteristics . This Graduation Design finishes a simple face recognition system based on PCA using the DM642 development board of TI .Mainly program to achieve the following functions: Video Capture ,Face Detection and Tracking , Image Denoising , the realization of PCA algorithm and Face Recognition. Now let me introduce the system’s design ideas: The video processing is based on a video capture box and a face tracking box. Only when the video capture box is inside the face tracking box , system collects a piece of image into memory . The face recognition system include two sections : the training stage and the recognition stage . In the training stage , system will collection a lot of images into memory , and train this pictures to form a feature space . Then make the faces projected onto the feature space . In the recognition stage , system will make the images which are going to be recognized projected onto the feature space .Get the face’s coordinate values in the feature space . Then we can recognize the face based on the Euclidean distance . Keywords: