基于改进多块局部二进制模式的人眼定位算法  

Human eye positioning algorithm based on improved multi-block local binary pattern

作　　者：鹿艳晶[1] 方加娟[1,2] 

机构地区：[1]郑州职业技术学院软件工程系,河南郑州450121 [2]南京理工大学计算机学院,江苏南京210094

出　　处：《现代电子技术》2017年第11期58-60,64,共4页Modern Electronics Technique

摘　　要：人眼定位是人脸识别系统中必须解决的问题,为了提高人眼定位的精确度和速度,提出一种基于改进多块局部二进制模式的眼睛中心定位算法。该算法构建了所有可能的弱分类器作为多分支的树枝,然后通过三个强分类器和迭代计算确认眼睛中心点。基于标准Bio ID和FERET图像数据库,将提出算法与其他两种方法进行比较,实验结果显示提出方法的精确度更高。失真图像的测试结果表明,对噪声失真而言,提出的算法足够稳健。The human eye positioning is a problem which must be solved for the human face recognition system. In order to improve the accuracy and speed of human eye localization,a human eye center positioning algorithm based on improved multiblock local binary pattern is proposed. All possible weak classifiers were constructed as the branchy branches with the algorithm.The center point of the human eye is confirmed by means of three strong classifiers and iterative computation. On the basis of standard Bio ID and FERET image database,the proposed algorithm is compared with other two methods. The experimental results show that the proposed algorithm has higher accuracy. The test results of the distorted image show that the algorithm is robust enough for noise distortion.

关 键 词：人脸识别 人眼中心定位 局部二进制模式 分类器 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]