基于图像特征的人眼定位学士学位论文.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 recognition is a very active area of research in recent years. Its applications range is very wide, such as identity authentication in safety system, video surveillance, target identification and tracking, as well as facial expression analysis, age analysis, lip reading and so on. compared with mouth and nose, Eyes are the most significant features of the face. which can provide more reliable, more important message, so eye detection is often necessary to dispose in face recognition. An algorithm for eyes location is presented in this paper based on maximum variance between two classes and gray- level integration projection. First, median filter is used to eliminate the noise, then the image blur caused by deficient exposal is cleared up using non - linear transform. Maximum variance between two classes is provided to get the binary image, and then the features are extracted from the image. Finally, by the way of gray-levelinte gration projection and human face configuration, we can easily find that the location of eyes is determined by the coordinate of the minimum in the diagram. Further feature detection can be done based on this result. KEYWORDS: Feature extraction; Eye location; Integration projection; Threshold segment 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 13 2.2.1 概述 13 2.2.2 图像噪声分类 13 2.2.3 图像系统噪声的特点 14 2.2.4 均值滤波 15 2.2.5 中值滤波 16 2.3 直方图均衡化 17 2.4 图像二值化 19 第三章 人眼定位算法 22 3.1 算法流程 22 3.2 人脸识别的常用方法 24 3.3 投影法原理 24 3.4 人眼左右边界的判定 25 3.5 眉眼区域的判定 26 3.6 眉眼区域的处理 27 3.