视频图像中的视觉疲劳实时检测方法研究-计算机工程与应用.PDF

Computer Engineering and Applications 计算机工程与应用 2012 ，48 （35） 147 视频图像中的视觉疲劳实时检测方法研究 兰 婷，普杰信 LAN Ting, PU Jiexin 河南科技大学 电子信息工程学院，河南 洛阳 471003 College of Electronic and Information Engineering, Henan University of Science and Technology, Luoyang, Henan 471003, China LAN Ting, PU Jiexin. Real time detection method of visual fatigue in video image. Computer Engineering and Applications, 2012, 48 （35 ）：147-150. Abstract ：It puts forward a new visual fatigue analysis algorithm which is applicable to detect face location and eyes state in driving at real time. The differential method is quickly used to find the target area of video. Face is posi- tioned by skin color segmentation combining YCbCr color space. The eyelid is detected by using the gray integral projection and Hough transform in the area of face. The detected data of eyelid is analyzed, and the condition of eyes is got. Value of EOD is to judge whether the driver is fatigue combined with blink analysis. The experimental result shows that the method can locate face quickly in complex background. The value of EOD is detected when the eyes open, and real-time test of visual fatigue is improved. Key words ：driving fatigue; video; skin color segmentation; Hough transform; blink cycle 摘 要：提出一种新的适用于驾驶中视觉疲劳实时检测的人脸定位及眼睛状态分析算法。采用差分法快速找 到视频图像中的目标运动区域，结合YCbCr 色彩空间进行肤色分割定位人脸。对脸部区域进行灰度积分投影 并结合Hough 变换检测眼睑。对检测到的眼睑进行数据分析，得到眼睛开闭情况，结合眨眼分析，获得EOD 值 来判断驾驶员是否疲劳。实验结果显示该方法能在复杂背景下快速定位人脸，检测到眼睛睁开时的EOD 值， 满足视觉疲劳检测的实时需要。 关键词：疲劳驾驶；视频；肤色分割；Hough 变换；眨眼周期 文献标识码：A 中图分类号：TP391 doi ：10.3778/j.issn. 1002-8331.1202-0172 1 引言 一些不足。模板匹配法需要预存多个模板，存储信息 近年来，随着机动车保有量的猛增，恶性交通事 量大，不易推广；Hough 变换圆检测方法定位准确但 [3] 故呈逐年上升趋势，采用一种非接触的、实时的疲劳 计算量较大；瞳孔检测则只能确定睁眼的情况