摘要
传统的获得人眼动姿态技术一般需要利用机械装置或是电极、镜片等直接与眼睛接触,给被试者的眼睛带来不适,具有很大的局限性。提出了基于图像处理方法的人眼动姿态识别技术。首先利用摄像机拍摄人脸图片,再利用水平差分和改进的垂直积分投影实现了人眼位置粗定位。在此基础上,采用了Harris角点检测方法寻找内外眼角,并提出基于Susan模板的差分极大值角点筛选方法来筛选出最佳内外眼角,最后提出了一种基于眼球低灰度值特征的分析方法确定眼球左右移动的姿态。提出的人眼动姿态识别技术具有定位准确率高、速度快、对人眼无伤害等优点,为人眼动姿态识别技术的应用提供了新的途径。
This paper proposes a technology based on image processing method to recognize the gesture of human eyes＇ movement.First of all,human face image is get from a camera.And then two methods which are called horizontal difference projection and improved vertical integral projection are used to locate human eyes roughly.On this basis,Harris corner detection method is adopted to find inner and outer canthus,and a picking method called the difference maximum angular point based on Susan template is put forward to screen the best inside and outside canthus.At last,an analysis method based on a low eye grey value is proposed to determine the gesture of eyes movement.
出处
《工业控制计算机》
2015年第3期31-33,共3页
Industrial Control Computer
关键词
人眼动特征
图像处理
人眼粗定位
角点检测
movement characteristics of human eyes
image processing
locate eyes roughly
corner detection
