一种人脸图象中瞳孔中心的提取方法  

A Method for Extracting Pupil Centers in Human Face Images

作　　者：许慰玲[1] 黄静霞[2] 

机构地区：[1]汕头大学广东省图象处理重点实验室,广东汕头515063 [2]汕头大学科研处,广东汕头515063

出　　处：《计算机工程与科学》2003年第4期18-22,共5页Computer Engineering & Science

摘　　要：本文提出一种人脸灰度图象中瞳孔中心精确提取的方法。首先采用积分投影法自动检测眼睛区域,然后利用灰度变化的强度、方向信息进行边缘提取和上、下眼睑轮廓跟踪,探讨基于眼睛几何特征的瞳孔中心精确定位方法。所采用的眼睑轮廓跟踪方法与传统的模板匹配法相比较,对眼睛特征的先验知识依赖程度较低,对不同类型的眼睛具有很好的适应性。所提出的瞳孔中心精确定位方法,与Hough变换的中心定位法相比较,计算量更小,且实验结果表明该方法具有很高的精确度。This paper proposes a method for the precise extraction of pupil centers from grayscale face images. First, the automatic detection of eye region using the integral projection method is discussed. Second, the extraction of the edge of the eye region and the tracking of the contour of eyelids based on a gradient magnitude and direction are analyzed. Finally, the precise location of pupil centers depending on eye geometry features are expanded. Compared with the traditional template matching method, the tracking of the eyelid contour depends less on the priori knowledge of eye features, and therefore the method is preferably suitable for different eye types. It can also be seen that the computing of the precise location method is much simpler compared with the iris center location method based on the Hough transform. Experimental results indicate that the proposed method provides higher precision.

关 键 词：人脸图象 瞳孔中心 模板匹配 边缘提取 人脸灰度图象 人脸识别 积分投影法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]