《基于几何特征人脸识别》由会员分享，可在线阅读，更多相关《基于几何特征人脸识别（11页珍藏版）》请在金锄头文库上搜索。
1、 基于几何特征的人脸识别基于几何特征的人脸识别 学号：学号： 姓姓 名：名： (上海大学 机电工程与自动化学院，200072) 摘摘 要要：人脸识别技术作为多学科领域的、具有挑战性的课题，它覆盖了数字图像处理、模式识别、神经 网络、数学等诸多学科的内容，同时也具有十分广泛的应用价值。在人脸识别领域，基于几何特征的人脸 识别算法因其计算简单、使用有效等特性，引起了人们的广泛注意，并已成为人脸图像特征提取和识别的 主流方法之一。本文定位人脸器官，通过人脸面部拓扑结构几何关系的先验知识，利用基于灰度投影的方 法在知识的层次上提取人脸面部主要器官特征，将人脸特征用一组几何特征向量表示，识别归结为特征向
2、 量之间的匹配。本文工作包括： (1)对灰度积分投影理论进行了详细的介绍和分析。这种方法是目前定位人脸轮廓的主要方法。在此基 础上对一种新的在竖直方向上定位人脸左右轮廓的灰度差投影法进行了改进。投影法本质上是一种基于统 计的方法，在具体应用时又结合了人脸特征分布的先验知识。这种方法不需要对积分投影图做任何平滑处 理等操作，因而算法简单，准确率高，速度很快。 (2)提出了一种精确定位眼睛的方法，该算法将眼区灰度总体分布特点与眼部灰度变化特点相结合，将 传统的积分投影法与灰度差累加值投影法相结合，通过大量试验选取合适的参数。试验结果表明，该算法 对光照变化不敏感，定位准确率高。运用灰度积分投影结合
3、人脸特征的先验知识定位鼻子，这种定位方法 得到的准确率也是比较高的。嘴巴的定位则利用投影法求得。 (3)本文根据定位出来的人脸器官选出七个特征点，即四个眼角点、鼻尖点和两个嘴角点。利用它们构 造了十个特征向量，并对其进行归一化运算。图像识别的最后一个过程就是分类，本文采取根据模式相似 性的最近距离分类器进行分类。用加权比值函数来计算特征相似度，更适用于人脸图像的识别和计算。如 何选择出合适的识别门限是个很复杂的问题，还有待于进一步研究。 关键词：关键词：人脸识别；灰度投影；几何特征；特征提取 The Research of Face Recognition Algorithm Based on
4、 Geometric Features Student number:15721637 Name:Zhao Pei-pei (Institute of electrical and mechanical engineering and automation, Shanghai University, Shanghai 200072, China) Abstract:the technology of human face recognition as a multi-disciplinary field and challenging psubject which contains digit
5、al image processing，pattern recognition，computer vision，neural network，psychology，physiology，mathematics and a good many subjectsIn the meantime，it also has widely usedIn the field of face recognition，the method of human face recognition based on geometric features has been paid great attention for
6、its simple calculation and availabilityAt present，it has become one of the dominant methods as the feature extraction and recognitionTMs article locates human face organs，through apriori knowledge of human face topological structure geometrical relationship，making use of method based on construct to
7、 extract the features of human face organs，expressing human face through a set of geometric feature vectorsThe recognition putting in summary is matched with feature vectorThis paper includes the following parts： (1)Have a detailed introduction and analysis about the theory of greyscale integrated p
8、rojectionThis method is now the main method of locating human faceWe put forward a new method called greyscale differential projection which is based on the previous method and locating the contour of human face vertical directlyProjection method is essentially based on statisticsIt combines the apr
9、iori knowledge of human face feature distribution in the applicationThis method neednt to do any pretreatment to the image and any smoothing treatment to the integrated projection imageSo this algorithm is simple；the accuracy is high; the speed is quick (2)Give an introduction about the method of lo
10、cating eyes preciselyThis algorithm combines the character of the eye area greyscale totally distribution and greyscale transformation；combines the methods of traditional integrated projection and differential projectionThe experiment led to the fact that this algorithm is not sensitive to the illum
11、ination transformation and has a high accuracyUsing greyscale integrated projection combines the apriori knowledge of human face character to locate noseThis location method also has high accuracyThe location of mouth is abtained through projection method (3)The choice of characteristic points needs
12、 enough information and cant go so far as to increase calculation quantityThis article chooses seven characteristic points，namely，four canthus points，tip of nose and two corners of mouth pointsConstruct ten eigenvectors using them and carries on the normalization calculation to themThe last process
13、of image recognition is classificationAfter adopting some standards to extract feature of human images，we construct category separability decision rule according to these characters and design classifierThis article takes use of minimum distance classification to classifyIt iS more suitable for huma
14、n face recognition and calculation using weighing ratio to calculate similarityHow to choose a suitable recognition threshold is a difficult problem and need further researchThis article ascertains it through a good many experiments Key Words：Face Recognition；Greyscale Projection；Geometric Character
15、s；Feature Extraction 1.1. 引言引言 1.11.1 人脸识别技术：人脸识别技术： 人脸识别是一个活跃的研究领域，是人类视觉最杰出的能力之一。虽然人脸识别的准确性要低于 虹膜、指纹的识别，但由于它的无侵害性和对用户最自然、最直观的方式，使人脸识别成为最容易被 接受的生物特征识别方式。 应用领域：人脸识别系统在金融、证券、社保、公安、军队及其他需要安全认证的行业和部门有 着广泛的应用。典型应用: 1）罪犯调查 3）重用门票 2）访问控制 4）信用卡 1.2 人体生物认证技术人体生物认证技术 人脸识别是人体生物认证技术的一种，人体生物的生物特征包括生理特征和行为特征两大类。
16、人体的生理特征主要包括人脸、指纹、掌纹、掌形、虹膜、视网膜、静脉、DNA、颅骨等，这 些特征是与生俱来的，是先天形成的； 而行为特征包括声纹、签名、步态、耳形、按键节奏、身体气味等，这些特征是由后天的生活 环境和生活习惯决定的。 这些生物特征本身固有的特点决定了其在生物认证中所起的作用是不同的。 生物特征识别: 人脸 脸部热量图 指纹 签名 Face Face heat figure Fingerprint Signature 图 1 Fig.1 常用生物特征的比较: 2. 人脸识别的过程人脸识别的过程 人脸识别过程主要通过三个步骤完成，即人脸检测、图像的预处理、面部特征提取和人脸对比识 别确认及分类器的设计，典型的人脸识别流程图如下： 人脸定位 输出结果 比对识别 特征提取 预处理 图像获取人脸检测 人脸特征 人脸库 图 2 人脸
