基于amShift算法及Kalman滤波的人脸跟踪(可编辑).docVIP

在计算机领域识别领域中，人脸的检测与跟踪是非常重要的一个课题。在人 脸识别、视频监控、智能安防等领域有着非常高的应用价值。随着信息技术的高 速发展，个人身份识别、智能人机交互等系统的应用需求，人脸检测与跟踪系具 有越来越广泛的实际应用价值和研究意义。 在本文中，人脸检测研究了 AdaBoost方法，将AdaBoost方法与肤色检验和 结合，去除竹景的干扰，不仅降低了误检率，还在一定程度上提高了检出率。人 脸跟踪研究了 CamShift方法，并针对CamShift方法存在的问题，如遇到遮挡物 干扰时跟踪失败、无法适应不同的运动速度等进行了改进；锁定目标时，对区域 内做肤色建模，然后引入Kalman滤波预测目标的位罝，最后引入特征匹配以进 一步约束CamShift的结果。实验证明改进后的方法能够很好的解决上述问题。 接下来根据本文所用方法,并利用OpenCV开发了一个人脸检测和跟踪系统。 利用该系统的实验结果可以发现，本文所提方法能够获得较好的人脸检测和跟踪 效果。 关键词：人脸检测；人脸跟踪；AdaBoost； CamShift ABSTRACT Human face detection and tracking is an important basic research project in the filed of Computer Vision and Pattern Recognition. It has great application value in the automatic human face recognition，video surveillance, smart safe-guard system and so on. With the rapid development of information technology，demand of personal identity recognition, intelligent human-computer interaction systems，human face detection and tracking has become widely used researched. In the article，we researched Adaboost method in human face detection，and by using complexion and Adaboost arithmetic can eliminate the noise signal effectively; it can both reduce the rate of false detections and improve the detection rate ? We researched CamShift method in human face tracking，and improved the CamShift method based on the shortcomings. In the first，modeling for complexion in the region when the target is locked. Second，use Kalman to predict the position of target. Last， introduce match-board to restrict the result of CamShift. Then develop a system of human face detection and tracking based on the discussed methods and OpenCV platform. By the simulation results of this system, we can find that these approaches can detect and track human face perfectly. Keywords:Human face detection； Human face tracking； AdBoost； CamShift TOC \o 1-5 \h \z 谢仑 5 1.1弓I胃 5 HYPERLINK \l bookmark3 \o Current Document \h 国内外研究现状 6 人月佥检测 6 1.2.2人脸跟踪 6 HYPERLINK \l bookmark4 \o Current Document \h 1.3论文内容安排 7 HYPERLINK \l bookmark5 \o Current D