基于眼动特征的机动车驾驶疲劳系统  

The Driver's Fatigue State of Detection System Based on Eyes' feature

作　　者：傅亚煌[1] 黄鸣宇[1] 

机构地区：[1]南京航空航天大学计算机科学与技术学院,江苏南京211100

出　　处：《电脑与信息技术》2013年第1期15-18,共4页Computer and Information Technology

基　　金：南京航空航天大学国家级大学生创新创业训练计划项目资助(项目编号:NO:201210287133)

摘　　要：本系统利用开源OpenCV图像处理函数库进行快速开发,总体设计方案包括如下几个方面:在图像预处理方面,系统将通过视频流获得的图像帧进行直方图均衡化,去除实际图像中存在的噪声,放大图像细节;在人眼识别与定位方面,系统采用成功的Adaboost算法,利用OpenCV提供的样本训练集和检测方法;人眼睁闭状态判断方面,系统利用自适应阈值化获取能够展示人眼完整形状的二值化图像,通过人眼区域内黑色像素点面积变化的比较来判断睁眼闭眼;最后,系统选取目前公认有效的PERCOLS疲劳程度评价指标对当前驾驶员的疲劳状态进行分析。The system is developed based on the function library of the open source OpenCV, the overall designing scheme is made up of five aspects, as the following shows： At the aspect of pretreatment of the images, the system will gain the every frames of video streaming for histogram equalization, and remove the actual existing noisy elements. At the aspect of face detection, the system successfully adopt the Adboost algorithm, the application of OpenCV provide sample of training set and testing method. At the aspect of the state of eyes judgment, the system use the adaptive thresholding acquisition to show the shape of the binary images of complete eyes, and then judging the state of your eyes through the change of eye areas with black pixels. Finally, the system choose the PERCOLS fatigue degree which is seen to be the most effective at present to judge and analyze the driver＇s fatigue condition.

关 键 词：OPENCV 人脸检测 人眼定位 二值化 PERCLOS 

分 类 号：TP317.4[自动化与计算机技术—计算机软件与理论]