[作业]OpenCV人脸识别介绍.doc

摘 要 人脸检测主要是基于计算机识别的一项数字化技术，用以准确获取人的脸部大小和位置信息，在进行人脸检测时，突出主要的脸部特征，淡化次要的环境、衣着等因素。对于某些情况下，人脸检测也可以计算出人脸，如眼睛，鼻子和嘴等精确的微妙特征。由于在安全检测系统，医学，档案管理，视频会议和人机交互等领域人脸检测系统都有光明的应用前景，因此人脸检测逐渐成为了两个跨学科领域研究的热门话题:人工智能和当前模式识别。本文基于OpenCV视觉库具体的设计并开发了对数字图像中的人脸检测的程序，所采用的人脸检测的原理主要是分类器训练模式(Adaboost算法)提取Haar特征的方法。它在整个软件极其重要的作用，图像中人脸的准确定位和识别都受图像处理好坏的直接影响。本次所设计的软件在图像处理部分所采用的方法是基于Adaboost算法进行Haar特征的提取，在此之上加以通过积分图方法来获取完整的级联分类器结构，进行人脸检测时，OpenCV级联分类器通过Adaboost人脸检测算法进行训练，此后采用不同情况下的实验样本完成精确定位以及检测试验。经过代码的设计和调试，在最后的测试中针对数字图像进行的人脸检测和定位达到了较好的效果，提高了定位和识别的正确率。 关键词：人脸检测，AdaBoost，分类器，OpenCV Abstract Face detection is mainly based on computer recognition of a digital technology, face size and location information to accurately obtain the person, during face detection, highlight the main facial features, dilute the secondary environment, clothing, and other factors . For some cases, face detection can also calculate a persons face, such as eyes, nose and mouth, and other subtle features accurate. Because in the field of human security detection systems, medical records management, video conferencing, and human-computer interaction face detection system has bright prospects, and therefore face detection is becoming a two interdisciplinary research fields hot topic: artificial intelligence and The current pattern recognition. This article is based. penCV vision library designed and developed specifically for digital image face detection process, the principles used face detection methods are mainly classifier training mode (Adaboost algorithm) to extract Haar features. It is in the vital role of the software, the image of the human face accurately locate and identify all that is good or bad a direct impact on the image processing. This software is designed image processing method used in part based Haar Adaboost algorithm to extract features, on top of this to be to get the full cascade classifier structure by integrating the diagram method for face detection, OpenCV cascade classifier is trained by Adaboost face