吉林大学学报(信息科学版) ›› 2016, Vol. 34 ›› Issue (2): 259-265.

• 论文 • 上一篇    下一篇

基于视频图像的眼动追踪系统算法

王际航, 刘富, 袁雨桐, 刘星   

  1. 吉林大学通信工程学院, 长春130022
  • 收稿日期:2015-12-17 出版日期:2016-03-24 发布日期:2016-06-17
  • 作者简介:王际航(1990—), 男, 长春人, 吉林大学硕士研究生, 主要从事模式识别与智能系统研究, (Tel)86-18626688973(E-mail)920161394@ qq. com; 通讯作者: 刘富(1968—), 男, 吉林农安人, 吉林大学教授, 博士生导师, 主要从事模式识别与智能系统研究, (Tel)86-13610708679(E-mail)liufu@ jlu. edu. cn。
  • 基金资助:

    国家自然科学基金资助项目(61503151)

Eye Tracking System Algorithm Based on Video Image Algorithm

WANG Jihang, LIU Fu, YUAN Yutong, LIU Xing   

  1. School of Telecommunication Engineering, Jilin University, Changchun 130022, China
  • Received:2015-12-17 Online:2016-03-24 Published:2016-06-17

摘要:

为了解决初步眼动追踪问题, 提出基于视频图像的实时性眼动追踪的快速算法。将RGB 色彩空间转换成YCbCr 空间, 利用肤色模型定位人脸。剪裁后, 用Sobel 算子边缘检测算法进行卷积处理, 对图像进行水平投影找到人眼大致位置, 对眼部进行粗定位。对该区域进行灰度投影, 分割左、右眼, 再分别对左、右眼进行定位, 从而得到人眼的精确定位。实验选取15 帧图片作为测试视频序列, 其结果表明, 该算法准确地解决了眼动追踪问题, 满足实时性要求。

关键词: 肤色模型, Sobel 算子, 灰度投影, 视频序列

Abstract:

In order to solve the problem of eye tracking, we put forward a fast real-time algorithm of eye tracking based on video image. Firstly,convert RGB color space to YCbCr space, and locate human faces by skin color model. After clipping, use Sobel operator edge detection algorithm in convolution processing, and then find the approximate location of eyes through horizontal gray-level projection to determine rough location of the eyes. Divide this area into left and right eye areas with the help of gray-scale projection,then respectively locate the left and right eye, finally get the precise localization of human eyes. Based on the algorithm, we got the results based on video sequence.

Key words: skin color model, Sobel operator, gray-level projection, video sequence

中图分类号: 

  • TP271