基于积分投影与区域生长的人眼定位方法  

Eyes Location Based on Integral Projection and Region Growth

作　　者：李德武 舒小华[1] 方盛 杨顺波 LI Dewu;SHU Xiaohua;FANG Sheng;YANG Shunbo(College of Electrical and Information Engineering,Hunan University of Technology,Zhuzhou Hunan 412007,China)

机构地区：[1]湖南工业大学电气与信息工程学院,湖南株洲412007

出　　处：《湖南工业大学学报》2018年第1期88-92,共5页Journal of Hunan University of Technology

基　　金：湖南工业大学教学研究基金资助项目(2015B17)

摘　　要：为了克服传统积分投影方法不能有效定位偏转人脸图像问题,提出一种基于积分投影与区域生长相结合的人眼定位方法。在定位人眼之前,通过肤色分割实现人脸粗定位,接着采用Otsu实现二值化、形态学操作以及区域标记算法等预处理操作。实验结果表明:采用的处理算法准确度较高,能为下一步的人眼状态分析打下良好的基础。In order to overcome the flaw commonly found in the traditional integration projection method that fails to detect the eyes in the deflected image,a new method has thus been proposed to locate the eyes based on the algorithms of integration projection and region growth.Before locating the human eye,a coarse location of faces can be achieved through skin color segmentation,followed by the realization of such preprocessing operations as binarization,morphological operation and region-labeling algorithm by adopting Otsu.The experiment shows that the processing algorithm is of high accuracy,thus laying a good foundation for the analysis of human eyes in the next step.

关 键 词：积分投影 区域生长 人眼定位 肤色分割 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]