基于肤色的人脸眼部区域图像自动分割方法  

Automatic segmentation method based on the color of the face eye area image

作　　者：刘祥楼[1,2] 段萍[1,2] 杨龙[1,2] 张明[1,2] 

机构地区：[1]东北石油大学,黑龙江大庆163318 [2]黑龙江省高校校企共建测试计量技术及仪器仪表工程研发中心,黑龙江大庆163318

出　　处：《自动化与仪器仪表》2013年第5期140-142,共3页Automation & Instrumentation

基　　金：黑龙江省研究生创新科研项目资助(YJSCX2011-121HLJ)

摘　　要：眼睛是人脸特征最主要的构成部分。基于眼部特征的人脸检索不仅可以提高人脸识别和检测的速度,而且能够降低识别算法的复杂度。要提取眼睛特征,首先要检测眼睛并提取眼睛区域图像,本文选择了一种基于肤色的人脸眼部区域图像自动分割方法。直接对彩色人脸图像利用YCbCr空间下肤色的色度与亮度的相关性将眼睛位置定位出来,再根据眼睛位置分割得到人脸眼部区域的带状图像。通过对图像库中的1000幅彩色人脸图像进行试验检测表明,计算机分割一张图像眼部区域的平均时间为1.785s,对图像库中人脸眼部区域图像分割提取的成功率为99.37%。The eyes are the most important components of the facial feature. Eye feature-based face retrieval can not only im- prove face recognition and detection speed, but also can reduce the complexity of the recognition algorithm. To extract the eye fea- ture, the first we must detect the eyes and extract the image of the eye region, selected based on the color of the face eye area auto- matic image segmentation method. Chroma and brightness of the skin tone color face image use of YCbCr space and eye position po- sitioning and ribbon image obtained by dividing the eye area of the face, according to the position of the eye. 1000 color face images in the image library assay showed that the computer extracted human face on the image library eye area image segmentation success rate of 99.37%.

关 键 词：眼部特征 区域分割 人眼定位 

分 类 号：TB114.1[理学—运筹学与控制论]