• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

J4 ›› 2014, Vol. 36 ›› Issue (04): 731-736.

• 论文 • 上一篇    下一篇

基于嘴巴特征点曲线拟合的哈欠检测

谢国波,陈云华,张灵,丁伍洋   

  1. (广东工业大学计算机学院,广东 广州 510006)
  • 收稿日期:2012-08-27 修回日期:2012-12-30 出版日期:2014-04-25 发布日期:2014-04-25
  • 基金资助:

    广东省教育部产学研项目(2012B091000058);广东省重点实验室建设专项项目(2011A091000046)

Yawning detection based on mouth feature point curve fitting         

XIE Guobo,CHEN Yunhua,ZHANG Ling,DING Wuyang   

  1. (College of Computer,Guangdong University of Technology,Guangzhou 510006,China)
  • Received:2012-08-27 Revised:2012-12-30 Online:2014-04-25 Published:2014-04-25

摘要:

针对疲劳分析中哈欠检测具有嘴角点定位困难、嘴巴张开大小及持续时间因人而异的特点,提出一种基于嘴巴内轮廓角点检测与曲线拟合的哈欠检测方法。首先利用角点检测获取嘴巴内轮廓上的若干点,对这些点进行曲线拟合建立嘴唇内轮廓数学模型;然后再对张口度曲线进行时间维度的分析,对哈欠进行二次判决。实验结果表明,该方法不仅能够更精确地获取开口度的大小,而且还能够降低哈欠的误检率。

关键词: 角点检测, 轮廓提取, 曲线拟合, 哈欠检测

Abstract:

In yawning detection for fatigue analysis,it is difficult to get the exact location of the two mouth corners,the open size and open duration varies for different individuals.For this problem, it proposes a yawning detection method based on mouth inner contour corner detection and curve fitting of those corner points. Firstly,several points are located on the contour of the mouth by corner detection.Secondly,the mathematical model of the mouth inner contour is established by curve fitting to those points.Thirdly,the mouth openness curve is analyzed in the time sequence to identify the yawning twice.Experimental results show that this method can obtain more precise mouth openness,and reduce false detection rate of yawning.

Key words: corner detection;contour extraction;curve fitting;yawning detection