基于图像特的征人眼定位.docVIP

内容摘要 人脸识别是?人类视觉最杰出的能力?之一，它的研究涉及模?式识别、图像处理、生?理，心理学、认知科学?，和基于其它生物特征?的身份鉴别方法以及计?算机人机感知学交互领?域都有密切联系其中人?眼的识别是计算机人脸?识别和智能监控中的重?要部分。本文所研究的?人眼识别对象都是针对?单人正面或半侧面图像?。该文提出了一种基于?最大类间方差阈值分割?和灰度积分投影技术的?眼睛定位方法。首先通?过图像预处理技术中的?中值滤波方法去除图像?噪声,并通过非线性变?换消除人脸图像因为曝?光条件不同而造成的模?糊,得到灰度分配较为?均匀的图像,然后利用?最大类间方差阈值法对?图像进行二值化处理,?将特征点从人脸图像分?割出来,并分别利用水?平和垂直灰度积分投影?曲线结合人脸的结构特?征找到眼睛的位置坐标?,实现了准确的眼睛定?位,从而为进一步提取?其它特征点打好了基础?。 关键词: 特?征提取；眼睛定位；积?分投影；阈值分割 ? ? Abstract? Computer ?face recog?nition is ?a very act?ive area o?f research? in recent? years. It?s applicat?ions range? is very w?ide, such ?as identit?y authenti?cation in ?safety sys?tem, video? surveilla?nce, targe?t identifi?cation and? tracking,? as well a?s facial e?xpression ?analysis, ?age analys?is, lip re?ading and ?so on. com?pared with? mouth and? nose, Eye?s are the ?most signi?ficant fea?tures of t?he face. w?hich can p?rovide mor?e reliable?, more imp?ortant mes?sage, so e?ye detecti?on is ofte?n necessar?y to dispo?se in face? recogniti?on. An alg?orithm for? eyes loca?tion is pr?esented in? this pape?r based on? maximum v?ariance be?tween two ?classes an?d gray- le?vel integr?ation proj?ection. Fi?rst, media?n filter i?s used to ?eliminate ?the noise,? then the ?image blur? caused by? deficient? exposal i?s cleared ?up using n?on - linea?r transfor?m. Maximum? variance ?between tw?o classes ?is provide?d to get t?he binary ?image, and? then the ?features a?re extract?ed from th?e image. F?inally, by? the way o?f gray-lev?elinte gra?tion proje?ction and ?human face? configura?tion, we c?an easily ?find that ?the locati?on of eyes? is determ?ined by th?e coordina?te of the ?minimum in? the diagr?am. Furthe?r feature ?detection ?can be don?e based on? this resu?lt. KEYWO?RDS: Feat?ure extrac?tion; Eye ?location; ?Integratio?n projecti?on; Thresh?old segmen?t ? 目 录 目 录 3? 第一章 绪 ?论 4 1.1 课题?的背景和意义 4 1?.2 论文的主要内容? 7 第二章 ?图像预处理基本知识 ?9 2.1 图像灰度?变换 9 2.1.1? 图像的灰度化 10? 2.1.2 图像灰?度求反 11 2.1?.3 图像灰度拉伸 ?12 2.2 图像平?滑去噪